* [PATCH net-next 0/2] net/mlx4: Add dynamic policy changes to the SRIOV eSwitch support
@ 2013-06-27 16:05 Or Gerlitz
2013-06-27 16:05 ` [PATCH net-next 1/2] net/mlx4_core: Dynamic VST to VST vlan/qos changes Or Gerlitz
` (2 more replies)
0 siblings, 3 replies; 5+ messages in thread
From: Or Gerlitz @ 2013-06-27 16:05 UTC (permalink / raw)
To: davem; +Cc: netdev, jackm, ronye, Or Gerlitz
Hi Dave,
This patch set from Jack and Rony uses the UPDATE_QP new firmware command
to allow for dynamic policy changes of PF controlled VF vlan/qos and link-state.
Done against net-next commit 8599b52e14 "bonding: add an option to fail when any
of arp_ip_target is inaccessible"
Or.
Jack Morgenstein (1):
net/mlx4_core: Dynamic VST to VST vlan/qos changes
Rony Efraim (1):
net/mlx4_core: Add HW enforcement to VF link state
drivers/net/ethernet/mellanox/mlx4/cmd.c | 149 +++++++++++++++++-
drivers/net/ethernet/mellanox/mlx4/fw.c | 5 +-
drivers/net/ethernet/mellanox/mlx4/mlx4.h | 21 +++
.../net/ethernet/mellanox/mlx4/resource_tracker.c | 163 +++++++++++++++++++-
include/linux/mlx4/cmd.h | 1 +
include/linux/mlx4/device.h | 3 +-
include/linux/mlx4/qp.h | 36 +++++
7 files changed, 362 insertions(+), 16 deletions(-)
^ permalink raw reply [flat|nested] 5+ messages in thread
* [PATCH net-next 1/2] net/mlx4_core: Dynamic VST to VST vlan/qos changes
2013-06-27 16:05 [PATCH net-next 0/2] net/mlx4: Add dynamic policy changes to the SRIOV eSwitch support Or Gerlitz
@ 2013-06-27 16:05 ` Or Gerlitz
2013-06-27 16:05 ` [PATCH net-next 2/2] net/mlx4_core: Add HW enforcement to VF link state Or Gerlitz
2013-07-01 20:12 ` [PATCH net-next 0/2] net/mlx4: Add dynamic policy changes to the SRIOV eSwitch support David Miller
2 siblings, 0 replies; 5+ messages in thread
From: Or Gerlitz @ 2013-06-27 16:05 UTC (permalink / raw)
To: davem; +Cc: netdev, jackm, ronye, Or Gerlitz
From: Jack Morgenstein <jackm@dev.mellanox.co.il>
Within VST mode, enable modifying the vlan and/or qos
for a VF without requiring unbind/rebind.
This requires firmware which supports the UPDATE_QP command.
(If the command is not available, we fall back to requiring
unbind/bind to activate these changes).
To avoid race conditions with modify-qp on QPs that are affected
by update-qp, this operation is performed on the comm_wq.
If the update operation succeeds for all the necessary QPs, a
vlan_unregister is performed for the abandoned vlan id.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/cmd.c | 122 ++++++++++++++++-
drivers/net/ethernet/mellanox/mlx4/fw.c | 5 +-
drivers/net/ethernet/mellanox/mlx4/mlx4.h | 20 +++
.../net/ethernet/mellanox/mlx4/resource_tracker.c | 145 +++++++++++++++++++-
include/linux/mlx4/cmd.h | 1 +
include/linux/mlx4/device.h | 3 +-
include/linux/mlx4/qp.h | 34 +++++
7 files changed, 319 insertions(+), 11 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
index df04c82..7b92789 100644
--- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
@@ -112,6 +112,14 @@ enum {
GO_BIT_TIMEOUT_MSECS = 10000
};
+enum mlx4_vlan_transition {
+ MLX4_VLAN_TRANSITION_VST_VST = 0,
+ MLX4_VLAN_TRANSITION_VST_VGT = 1,
+ MLX4_VLAN_TRANSITION_VGT_VST = 2,
+ MLX4_VLAN_TRANSITION_VGT_VGT = 3,
+};
+
+
struct mlx4_cmd_context {
struct completion done;
int result;
@@ -792,6 +800,15 @@ static int mlx4_MAD_IFC_wrapper(struct mlx4_dev *dev, int slave,
vhcr->op, MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE);
}
+int MLX4_CMD_UPDATE_QP_wrapper(struct mlx4_dev *dev, int slave,
+ struct mlx4_vhcr *vhcr,
+ struct mlx4_cmd_mailbox *inbox,
+ struct mlx4_cmd_mailbox *outbox,
+ struct mlx4_cmd_info *cmd)
+{
+ return -EPERM;
+}
+
int mlx4_DMA_wrapper(struct mlx4_dev *dev, int slave,
struct mlx4_vhcr *vhcr,
struct mlx4_cmd_mailbox *inbox,
@@ -1226,6 +1243,15 @@ static struct mlx4_cmd_info cmd_info[] = {
.wrapper = mlx4_GEN_QP_wrapper
},
{
+ .opcode = MLX4_CMD_UPDATE_QP,
+ .has_inbox = false,
+ .has_outbox = false,
+ .out_is_imm = false,
+ .encode_slave_id = false,
+ .verify = NULL,
+ .wrapper = MLX4_CMD_UPDATE_QP_wrapper
+ },
+ {
.opcode = MLX4_CMD_CONF_SPECIAL_QP,
.has_inbox = false,
.has_outbox = false,
@@ -1495,6 +1521,72 @@ out:
return ret;
}
+
+int mlx4_master_immediate_activate_vlan_qos(struct mlx4_priv *priv,
+ int slave, int port)
+{
+ struct mlx4_vport_oper_state *vp_oper;
+ struct mlx4_vport_state *vp_admin;
+ struct mlx4_vf_immed_vlan_work *work;
+ int err;
+ int admin_vlan_ix = NO_INDX;
+
+ vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
+ vp_admin = &priv->mfunc.master.vf_admin[slave].vport[port];
+
+ if (vp_oper->state.default_vlan == vp_admin->default_vlan &&
+ vp_oper->state.default_qos == vp_admin->default_qos)
+ return 0;
+
+ work = kzalloc(sizeof(*work), GFP_KERNEL);
+ if (!work)
+ return -ENOMEM;
+
+ if (vp_oper->state.default_vlan != vp_admin->default_vlan) {
+ err = __mlx4_register_vlan(&priv->dev, port,
+ vp_admin->default_vlan,
+ &admin_vlan_ix);
+ if (err) {
+ mlx4_warn((&priv->dev),
+ "No vlan resources slave %d, port %d\n",
+ slave, port);
+ return err;
+ }
+ work->flags |= MLX4_VF_IMMED_VLAN_FLAG_VLAN;
+ mlx4_dbg((&(priv->dev)),
+ "alloc vlan %d idx %d slave %d port %d\n",
+ (int)(vp_admin->default_vlan),
+ admin_vlan_ix, slave, port);
+ }
+
+ /* save original vlan ix and vlan id */
+ work->orig_vlan_id = vp_oper->state.default_vlan;
+ work->orig_vlan_ix = vp_oper->vlan_idx;
+
+ /* handle new qos */
+ if (vp_oper->state.default_qos != vp_admin->default_qos)
+ work->flags |= MLX4_VF_IMMED_VLAN_FLAG_QOS;
+
+ if (work->flags & MLX4_VF_IMMED_VLAN_FLAG_VLAN)
+ vp_oper->vlan_idx = admin_vlan_ix;
+
+ vp_oper->state.default_vlan = vp_admin->default_vlan;
+ vp_oper->state.default_qos = vp_admin->default_qos;
+
+ /* iterate over QPs owned by this slave, using UPDATE_QP */
+ work->port = port;
+ work->slave = slave;
+ work->qos = vp_oper->state.default_qos;
+ work->vlan_id = vp_oper->state.default_vlan;
+ work->vlan_ix = vp_oper->vlan_idx;
+ work->priv = priv;
+ INIT_WORK(&work->work, mlx4_vf_immed_vlan_work_handler);
+ queue_work(priv->mfunc.master.comm_wq, &work->work);
+
+ return 0;
+}
+
+
static int mlx4_master_activate_admin_state(struct mlx4_priv *priv, int slave)
{
int port, err;
@@ -2109,11 +2201,18 @@ int mlx4_set_vf_mac(struct mlx4_dev *dev, int port, int vf, u64 mac)
}
EXPORT_SYMBOL_GPL(mlx4_set_vf_mac);
+static int calculate_transition(u16 oper_vlan, u16 admin_vlan)
+{
+ return (2 * (oper_vlan == MLX4_VGT) + (admin_vlan == MLX4_VGT));
+}
+
int mlx4_set_vf_vlan(struct mlx4_dev *dev, int port, int vf, u16 vlan, u8 qos)
{
struct mlx4_priv *priv = mlx4_priv(dev);
- struct mlx4_vport_state *s_info;
+ struct mlx4_vport_oper_state *vf_oper;
+ struct mlx4_vport_state *vf_admin;
int slave;
+ enum mlx4_vlan_transition vlan_trans;
if ((!mlx4_is_master(dev)) ||
!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_VLAN_CONTROL))
@@ -2126,12 +2225,25 @@ int mlx4_set_vf_vlan(struct mlx4_dev *dev, int port, int vf, u16 vlan, u8 qos)
if (slave < 0)
return -EINVAL;
- s_info = &priv->mfunc.master.vf_admin[slave].vport[port];
+ vf_admin = &priv->mfunc.master.vf_admin[slave].vport[port];
+ vf_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
+
if ((0 == vlan) && (0 == qos))
- s_info->default_vlan = MLX4_VGT;
+ vf_admin->default_vlan = MLX4_VGT;
else
- s_info->default_vlan = vlan;
- s_info->default_qos = qos;
+ vf_admin->default_vlan = vlan;
+ vf_admin->default_qos = qos;
+
+ vlan_trans = calculate_transition(vf_oper->state.default_vlan,
+ vf_admin->default_vlan);
+
+ if (priv->mfunc.master.slave_state[slave].active &&
+ dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_UPDATE_QP &&
+ vlan_trans == MLX4_VLAN_TRANSITION_VST_VST) {
+ mlx4_info(dev, "updating vf %d port %d config params immediately\n",
+ vf, port);
+ mlx4_master_immediate_activate_vlan_qos(priv, slave, port);
+ }
return 0;
}
EXPORT_SYMBOL_GPL(mlx4_set_vf_vlan);
diff --git a/drivers/net/ethernet/mellanox/mlx4/fw.c b/drivers/net/ethernet/mellanox/mlx4/fw.c
index 569bbe3..8873d68 100644
--- a/drivers/net/ethernet/mellanox/mlx4/fw.c
+++ b/drivers/net/ethernet/mellanox/mlx4/fw.c
@@ -133,7 +133,8 @@ static void dump_dev_cap_flags2(struct mlx4_dev *dev, u64 flags)
[4] = "Automatic MAC reassignment support",
[5] = "Time stamping support",
[6] = "VST (control vlan insertion/stripping) support",
- [7] = "FSM (MAC anti-spoofing) support"
+ [7] = "FSM (MAC anti-spoofing) support",
+ [8] = "Dynamic QP updates support"
};
int i;
@@ -659,6 +660,8 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
QUERY_DEV_CAP_MAX_COUNTERS_OFFSET);
MLX4_GET(field32, outbox, QUERY_DEV_CAP_EXT_2_FLAGS_OFFSET);
+ if (field32 & (1 << 16))
+ dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_UPDATE_QP;
if (field32 & (1 << 26))
dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_VLAN_CONTROL;
if (field32 & (1 << 20))
diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4.h b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
index 7527293..5abcb65 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mlx4.h
+++ b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
@@ -571,6 +571,24 @@ struct mlx4_cmd {
u8 comm_toggle;
};
+enum {
+ MLX4_VF_IMMED_VLAN_FLAG_VLAN = 1 << 0,
+ MLX4_VF_IMMED_VLAN_FLAG_QOS = 1 << 1,
+};
+struct mlx4_vf_immed_vlan_work {
+ struct work_struct work;
+ struct mlx4_priv *priv;
+ int flags;
+ int slave;
+ int vlan_ix;
+ int orig_vlan_ix;
+ u8 port;
+ u8 qos;
+ u16 vlan_id;
+ u16 orig_vlan_id;
+};
+
+
struct mlx4_uar_table {
struct mlx4_bitmap bitmap;
};
@@ -1218,4 +1236,6 @@ static inline spinlock_t *mlx4_tlock(struct mlx4_dev *dev)
#define NOT_MASKED_PD_BITS 17
+void mlx4_vf_immed_vlan_work_handler(struct work_struct *_work);
+
#endif /* MLX4_H */
diff --git a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
index 1157f02..46323a2 100644
--- a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
+++ b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
@@ -101,6 +101,8 @@ struct res_qp {
spinlock_t mcg_spl;
int local_qpn;
atomic_t ref_count;
+ u32 qpc_flags;
+ u8 sched_queue;
};
enum res_mtt_states {
@@ -355,7 +357,7 @@ static void update_gid(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *inbox,
static int update_vport_qp_param(struct mlx4_dev *dev,
struct mlx4_cmd_mailbox *inbox,
- u8 slave)
+ u8 slave, u32 qpn)
{
struct mlx4_qp_context *qpc = inbox->buf + 8;
struct mlx4_vport_oper_state *vp_oper;
@@ -369,9 +371,17 @@ static int update_vport_qp_param(struct mlx4_dev *dev,
if (MLX4_VGT != vp_oper->state.default_vlan) {
qp_type = (be32_to_cpu(qpc->flags) >> 16) & 0xff;
- if (MLX4_QP_ST_RC == qp_type)
+ if (MLX4_QP_ST_RC == qp_type ||
+ (MLX4_QP_ST_UD == qp_type &&
+ !mlx4_is_qp_reserved(dev, qpn)))
return -EINVAL;
+ /* the reserved QPs (special, proxy, tunnel)
+ * do not operate over vlans
+ */
+ if (mlx4_is_qp_reserved(dev, qpn))
+ return 0;
+
/* force strip vlan by clear vsd */
qpc->param3 &= ~cpu_to_be32(MLX4_STRIP_VLAN);
if (0 != vp_oper->state.default_vlan) {
@@ -2114,6 +2124,8 @@ int mlx4_RST2INIT_QP_wrapper(struct mlx4_dev *dev, int slave,
if (err)
return err;
qp->local_qpn = local_qpn;
+ qp->sched_queue = 0;
+ qp->qpc_flags = be32_to_cpu(qpc->flags);
err = get_res(dev, slave, mtt_base, RES_MTT, &mtt);
if (err)
@@ -2836,6 +2848,9 @@ int mlx4_INIT2RTR_QP_wrapper(struct mlx4_dev *dev, int slave,
{
int err;
struct mlx4_qp_context *qpc = inbox->buf + 8;
+ int qpn = vhcr->in_modifier & 0x7fffff;
+ struct res_qp *qp;
+ u8 orig_sched_queue;
err = verify_qp_parameters(dev, inbox, QP_TRANS_INIT2RTR, slave);
if (err)
@@ -2844,11 +2859,30 @@ int mlx4_INIT2RTR_QP_wrapper(struct mlx4_dev *dev, int slave,
update_pkey_index(dev, slave, inbox);
update_gid(dev, inbox, (u8)slave);
adjust_proxy_tun_qkey(dev, vhcr, qpc);
- err = update_vport_qp_param(dev, inbox, slave);
+ orig_sched_queue = qpc->pri_path.sched_queue;
+ err = update_vport_qp_param(dev, inbox, slave, qpn);
if (err)
return err;
- return mlx4_GEN_QP_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+ err = get_res(dev, slave, qpn, RES_QP, &qp);
+ if (err)
+ return err;
+ if (qp->com.from_state != RES_QP_HW) {
+ err = -EBUSY;
+ goto out;
+ }
+
+ err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+out:
+ /* if no error, save sched queue value passed in by VF. This is
+ * essentially the QOS value provided by the VF. This will be useful
+ * if we allow dynamic changes from VST back to VGT
+ */
+ if (!err)
+ qp->sched_queue = orig_sched_queue;
+
+ put_res(dev, slave, qpn, RES_QP);
+ return err;
}
int mlx4_RTR2RTS_QP_wrapper(struct mlx4_dev *dev, int slave,
@@ -3932,3 +3966,106 @@ void mlx4_delete_all_resources_for_slave(struct mlx4_dev *dev, int slave)
rem_slave_xrcdns(dev, slave);
mutex_unlock(&priv->mfunc.master.res_tracker.slave_list[slave].mutex);
}
+
+void mlx4_vf_immed_vlan_work_handler(struct work_struct *_work)
+{
+ struct mlx4_vf_immed_vlan_work *work =
+ container_of(_work, struct mlx4_vf_immed_vlan_work, work);
+ struct mlx4_cmd_mailbox *mailbox;
+ struct mlx4_update_qp_context *upd_context;
+ struct mlx4_dev *dev = &work->priv->dev;
+ struct mlx4_resource_tracker *tracker =
+ &work->priv->mfunc.master.res_tracker;
+ struct list_head *qp_list =
+ &tracker->slave_list[work->slave].res_list[RES_QP];
+ struct res_qp *qp;
+ struct res_qp *tmp;
+ u64 qp_mask = ((1ULL << MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_UNTAGGED) |
+ (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_1P) |
+ (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_TAGGED) |
+ (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_UNTAGGED) |
+ (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_1P) |
+ (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_TAGGED) |
+ (1ULL << MLX4_UPD_QP_PATH_MASK_VLAN_INDEX) |
+ (1ULL << MLX4_UPD_QP_PATH_MASK_SCHED_QUEUE));
+
+ int err;
+ int port, errors = 0;
+ u8 vlan_control;
+
+ if (mlx4_is_slave(dev)) {
+ mlx4_warn(dev, "Trying to update-qp in slave %d\n",
+ work->slave);
+ goto out;
+ }
+
+ mailbox = mlx4_alloc_cmd_mailbox(dev);
+ if (IS_ERR(mailbox))
+ goto out;
+
+ if (!work->vlan_id)
+ vlan_control = MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
+ MLX4_VLAN_CTRL_ETH_RX_BLOCK_TAGGED;
+ else
+ vlan_control = MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
+ MLX4_VLAN_CTRL_ETH_RX_BLOCK_PRIO_TAGGED |
+ MLX4_VLAN_CTRL_ETH_RX_BLOCK_UNTAGGED;
+
+ upd_context = mailbox->buf;
+ upd_context->primary_addr_path_mask = cpu_to_be64(qp_mask);
+ upd_context->qp_context.pri_path.vlan_control = vlan_control;
+ upd_context->qp_context.pri_path.vlan_index = work->vlan_ix;
+
+ spin_lock_irq(mlx4_tlock(dev));
+ list_for_each_entry_safe(qp, tmp, qp_list, com.list) {
+ spin_unlock_irq(mlx4_tlock(dev));
+ if (qp->com.owner == work->slave) {
+ if (qp->com.from_state != RES_QP_HW ||
+ !qp->sched_queue || /* no INIT2RTR trans yet */
+ mlx4_is_qp_reserved(dev, qp->local_qpn) ||
+ qp->qpc_flags & (1 << MLX4_RSS_QPC_FLAG_OFFSET)) {
+ spin_lock_irq(mlx4_tlock(dev));
+ continue;
+ }
+ port = (qp->sched_queue >> 6 & 1) + 1;
+ if (port != work->port) {
+ spin_lock_irq(mlx4_tlock(dev));
+ continue;
+ }
+ upd_context->qp_context.pri_path.sched_queue =
+ qp->sched_queue & 0xC7;
+ upd_context->qp_context.pri_path.sched_queue |=
+ ((work->qos & 0x7) << 3);
+
+ err = mlx4_cmd(dev, mailbox->dma,
+ qp->local_qpn & 0xffffff,
+ 0, MLX4_CMD_UPDATE_QP,
+ MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE);
+ if (err) {
+ mlx4_info(dev, "UPDATE_QP failed for slave %d, "
+ "port %d, qpn %d (%d)\n",
+ work->slave, port, qp->local_qpn,
+ err);
+ errors++;
+ }
+ }
+ spin_lock_irq(mlx4_tlock(dev));
+ }
+ spin_unlock_irq(mlx4_tlock(dev));
+ mlx4_free_cmd_mailbox(dev, mailbox);
+
+ if (errors)
+ mlx4_err(dev, "%d UPDATE_QP failures for slave %d, port %d\n",
+ errors, work->slave, work->port);
+
+ /* unregister previous vlan_id if needed and we had no errors
+ * while updating the QPs
+ */
+ if (work->flags & MLX4_VF_IMMED_VLAN_FLAG_VLAN && !errors &&
+ NO_INDX != work->orig_vlan_ix)
+ __mlx4_unregister_vlan(&work->priv->dev, work->port,
+ work->orig_vlan_ix);
+out:
+ kfree(work);
+ return;
+}
diff --git a/include/linux/mlx4/cmd.h b/include/linux/mlx4/cmd.h
index 8074a97..bb1c809 100644
--- a/include/linux/mlx4/cmd.h
+++ b/include/linux/mlx4/cmd.h
@@ -111,6 +111,7 @@ enum {
MLX4_CMD_INIT2INIT_QP = 0x2d,
MLX4_CMD_SUSPEND_QP = 0x32,
MLX4_CMD_UNSUSPEND_QP = 0x33,
+ MLX4_CMD_UPDATE_QP = 0x61,
/* special QP and management commands */
MLX4_CMD_CONF_SPECIAL_QP = 0x23,
MLX4_CMD_MAD_IFC = 0x24,
diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
index a51b013..52c23a8 100644
--- a/include/linux/mlx4/device.h
+++ b/include/linux/mlx4/device.h
@@ -157,7 +157,8 @@ enum {
MLX4_DEV_CAP_FLAGS2_REASSIGN_MAC_EN = 1LL << 4,
MLX4_DEV_CAP_FLAG2_TS = 1LL << 5,
MLX4_DEV_CAP_FLAG2_VLAN_CONTROL = 1LL << 6,
- MLX4_DEV_CAP_FLAG2_FSM = 1LL << 7
+ MLX4_DEV_CAP_FLAG2_FSM = 1LL << 7,
+ MLX4_DEV_CAP_FLAG2_UPDATE_QP = 1LL << 8
};
enum {
diff --git a/include/linux/mlx4/qp.h b/include/linux/mlx4/qp.h
index 352eec9..f43e32a 100644
--- a/include/linux/mlx4/qp.h
+++ b/include/linux/mlx4/qp.h
@@ -206,6 +206,40 @@ struct mlx4_qp_context {
u32 reserved5[10];
};
+struct mlx4_update_qp_context {
+ __be64 qp_mask;
+ __be64 primary_addr_path_mask;
+ __be64 secondary_addr_path_mask;
+ u64 reserved1;
+ struct mlx4_qp_context qp_context;
+ u64 reserved2[58];
+};
+
+enum {
+ MLX4_UPD_QP_MASK_PM_STATE = 32,
+ MLX4_UPD_QP_MASK_VSD = 33,
+};
+
+enum {
+ MLX4_UPD_QP_PATH_MASK_PKEY_INDEX = 0 + 32,
+ MLX4_UPD_QP_PATH_MASK_FSM = 1 + 32,
+ MLX4_UPD_QP_PATH_MASK_MAC_INDEX = 2 + 32,
+ MLX4_UPD_QP_PATH_MASK_FVL = 3 + 32,
+ MLX4_UPD_QP_PATH_MASK_CV = 4 + 32,
+ MLX4_UPD_QP_PATH_MASK_VLAN_INDEX = 5 + 32,
+ MLX4_UPD_QP_PATH_MASK_ETH_HIDE_CQE_VLAN = 6 + 32,
+ MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_UNTAGGED = 7 + 32,
+ MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_1P = 8 + 32,
+ MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_TAGGED = 9 + 32,
+ MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_UNTAGGED = 10 + 32,
+ MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_1P = 11 + 32,
+ MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_TAGGED = 12 + 32,
+ MLX4_UPD_QP_PATH_MASK_FEUP = 13 + 32,
+ MLX4_UPD_QP_PATH_MASK_SCHED_QUEUE = 14 + 32,
+ MLX4_UPD_QP_PATH_MASK_IF_COUNTER_INDEX = 15 + 32,
+ MLX4_UPD_QP_PATH_MASK_FVL_RX = 16 + 32,
+};
+
enum { /* param3 */
MLX4_STRIP_VLAN = 1 << 30
};
--
1.7.1
^ permalink raw reply related [flat|nested] 5+ messages in thread
* [PATCH net-next 2/2] net/mlx4_core: Add HW enforcement to VF link state
2013-06-27 16:05 [PATCH net-next 0/2] net/mlx4: Add dynamic policy changes to the SRIOV eSwitch support Or Gerlitz
2013-06-27 16:05 ` [PATCH net-next 1/2] net/mlx4_core: Dynamic VST to VST vlan/qos changes Or Gerlitz
@ 2013-06-27 16:05 ` Or Gerlitz
2013-07-01 20:12 ` [PATCH net-next 0/2] net/mlx4: Add dynamic policy changes to the SRIOV eSwitch support David Miller
2 siblings, 0 replies; 5+ messages in thread
From: Or Gerlitz @ 2013-06-27 16:05 UTC (permalink / raw)
To: davem; +Cc: netdev, jackm, ronye, Or Gerlitz
From: Rony Efraim <ronye@mellanox.com>
When the firmware supports the UPDATE_QP command, if the VF link is disabled,
block all QPs opened by the VF, by programming the UPDATE_QP command to drop
all RX & TX traffic to/from these QPs. Operates only in VST mode.
Signed-off-by: Rony Efraim <ronye@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/cmd.c | 57 +++++++++++++-------
drivers/net/ethernet/mellanox/mlx4/mlx4.h | 1 +
.../net/ethernet/mellanox/mlx4/resource_tracker.c | 22 +++++++-
include/linux/mlx4/qp.h | 2 +
4 files changed, 60 insertions(+), 22 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
index 7b92789..9bf6ec1 100644
--- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
@@ -1521,6 +1521,10 @@ out:
return ret;
}
+static int calculate_transition(u16 oper_vlan, u16 admin_vlan)
+{
+ return (2 * (oper_vlan == MLX4_VGT) + (admin_vlan == MLX4_VGT));
+}
int mlx4_master_immediate_activate_vlan_qos(struct mlx4_priv *priv,
int slave, int port)
@@ -1528,16 +1532,37 @@ int mlx4_master_immediate_activate_vlan_qos(struct mlx4_priv *priv,
struct mlx4_vport_oper_state *vp_oper;
struct mlx4_vport_state *vp_admin;
struct mlx4_vf_immed_vlan_work *work;
+ struct mlx4_dev *dev = &(priv->dev);
int err;
int admin_vlan_ix = NO_INDX;
+ enum mlx4_vlan_transition vlan_trans;
vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
vp_admin = &priv->mfunc.master.vf_admin[slave].vport[port];
if (vp_oper->state.default_vlan == vp_admin->default_vlan &&
- vp_oper->state.default_qos == vp_admin->default_qos)
+ vp_oper->state.default_qos == vp_admin->default_qos &&
+ vp_oper->state.link_state == vp_admin->link_state)
return 0;
+ vlan_trans = calculate_transition(vp_oper->state.default_vlan,
+ vp_admin->default_vlan);
+
+ if (!(priv->mfunc.master.slave_state[slave].active &&
+ dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_UPDATE_QP &&
+ vlan_trans == MLX4_VLAN_TRANSITION_VST_VST)) {
+ /* even if the UPDATE_QP command isn't supported, we still want
+ * to set this VF link according to the admin directive
+ */
+ vp_oper->state.link_state = vp_admin->link_state;
+ return -1;
+ }
+
+ mlx4_dbg(dev, "updating immediately admin params slave %d port %d\n",
+ slave, port);
+ mlx4_dbg(dev, "vlan %d QoS %d link down %d\n", vp_admin->default_vlan,
+ vp_admin->default_qos, vp_admin->link_state);
+
work = kzalloc(sizeof(*work), GFP_KERNEL);
if (!work)
return -ENOMEM;
@@ -1572,6 +1597,10 @@ int mlx4_master_immediate_activate_vlan_qos(struct mlx4_priv *priv,
vp_oper->state.default_vlan = vp_admin->default_vlan;
vp_oper->state.default_qos = vp_admin->default_qos;
+ vp_oper->state.link_state = vp_admin->link_state;
+
+ if (vp_admin->link_state == IFLA_VF_LINK_STATE_DISABLE)
+ work->flags |= MLX4_VF_IMMED_VLAN_FLAG_LINK_DISABLE;
/* iterate over QPs owned by this slave, using UPDATE_QP */
work->port = port;
@@ -2201,10 +2230,6 @@ int mlx4_set_vf_mac(struct mlx4_dev *dev, int port, int vf, u64 mac)
}
EXPORT_SYMBOL_GPL(mlx4_set_vf_mac);
-static int calculate_transition(u16 oper_vlan, u16 admin_vlan)
-{
- return (2 * (oper_vlan == MLX4_VGT) + (admin_vlan == MLX4_VGT));
-}
int mlx4_set_vf_vlan(struct mlx4_dev *dev, int port, int vf, u16 vlan, u8 qos)
{
@@ -2212,7 +2237,6 @@ int mlx4_set_vf_vlan(struct mlx4_dev *dev, int port, int vf, u16 vlan, u8 qos)
struct mlx4_vport_oper_state *vf_oper;
struct mlx4_vport_state *vf_admin;
int slave;
- enum mlx4_vlan_transition vlan_trans;
if ((!mlx4_is_master(dev)) ||
!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_VLAN_CONTROL))
@@ -2234,16 +2258,10 @@ int mlx4_set_vf_vlan(struct mlx4_dev *dev, int port, int vf, u16 vlan, u8 qos)
vf_admin->default_vlan = vlan;
vf_admin->default_qos = qos;
- vlan_trans = calculate_transition(vf_oper->state.default_vlan,
- vf_admin->default_vlan);
-
- if (priv->mfunc.master.slave_state[slave].active &&
- dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_UPDATE_QP &&
- vlan_trans == MLX4_VLAN_TRANSITION_VST_VST) {
- mlx4_info(dev, "updating vf %d port %d config params immediately\n",
+ if (mlx4_master_immediate_activate_vlan_qos(priv, slave, port))
+ mlx4_info(dev,
+ "updating vf %d port %d config will take effect on next VF restart\n",
vf, port);
- mlx4_master_immediate_activate_vlan_qos(priv, slave, port);
- }
return 0;
}
EXPORT_SYMBOL_GPL(mlx4_set_vf_vlan);
@@ -2307,7 +2325,6 @@ int mlx4_set_vf_link_state(struct mlx4_dev *dev, int port, int vf, int link_stat
{
struct mlx4_priv *priv = mlx4_priv(dev);
struct mlx4_vport_state *s_info;
- struct mlx4_vport_oper_state *vp_oper;
int slave;
u8 link_stat_event;
@@ -2337,14 +2354,16 @@ int mlx4_set_vf_link_state(struct mlx4_dev *dev, int port, int vf, int link_stat
link_state, slave, port);
return -EINVAL;
};
- /* update the admin & oper state on the link state */
s_info = &priv->mfunc.master.vf_admin[slave].vport[port];
- vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
s_info->link_state = link_state;
- vp_oper->state.link_state = link_state;
/* send event */
mlx4_gen_port_state_change_eqe(dev, slave, port, link_stat_event);
+
+ if (mlx4_master_immediate_activate_vlan_qos(priv, slave, port))
+ mlx4_dbg(dev,
+ "updating vf %d port %d no link state HW enforcment\n",
+ vf, port);
return 0;
}
EXPORT_SYMBOL_GPL(mlx4_set_vf_link_state);
diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4.h b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
index 5abcb65..17d9277 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mlx4.h
+++ b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
@@ -574,6 +574,7 @@ struct mlx4_cmd {
enum {
MLX4_VF_IMMED_VLAN_FLAG_VLAN = 1 << 0,
MLX4_VF_IMMED_VLAN_FLAG_QOS = 1 << 1,
+ MLX4_VF_IMMED_VLAN_FLAG_LINK_DISABLE = 1 << 2,
};
struct mlx4_vf_immed_vlan_work {
struct work_struct work;
diff --git a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
index 46323a2..f984a89 100644
--- a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
+++ b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
@@ -384,7 +384,17 @@ static int update_vport_qp_param(struct mlx4_dev *dev,
/* force strip vlan by clear vsd */
qpc->param3 &= ~cpu_to_be32(MLX4_STRIP_VLAN);
- if (0 != vp_oper->state.default_vlan) {
+
+ if (vp_oper->state.link_state == IFLA_VF_LINK_STATE_DISABLE &&
+ dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_UPDATE_QP) {
+ qpc->pri_path.vlan_control =
+ MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
+ MLX4_VLAN_CTRL_ETH_TX_BLOCK_PRIO_TAGGED |
+ MLX4_VLAN_CTRL_ETH_TX_BLOCK_UNTAGGED |
+ MLX4_VLAN_CTRL_ETH_RX_BLOCK_PRIO_TAGGED |
+ MLX4_VLAN_CTRL_ETH_RX_BLOCK_UNTAGGED |
+ MLX4_VLAN_CTRL_ETH_RX_BLOCK_TAGGED;
+ } else if (0 != vp_oper->state.default_vlan) {
qpc->pri_path.vlan_control =
MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
MLX4_VLAN_CTRL_ETH_RX_BLOCK_PRIO_TAGGED |
@@ -4002,8 +4012,14 @@ void mlx4_vf_immed_vlan_work_handler(struct work_struct *_work)
mailbox = mlx4_alloc_cmd_mailbox(dev);
if (IS_ERR(mailbox))
goto out;
-
- if (!work->vlan_id)
+ if (work->flags & MLX4_VF_IMMED_VLAN_FLAG_LINK_DISABLE) /* block all */
+ vlan_control = MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
+ MLX4_VLAN_CTRL_ETH_TX_BLOCK_PRIO_TAGGED |
+ MLX4_VLAN_CTRL_ETH_TX_BLOCK_UNTAGGED |
+ MLX4_VLAN_CTRL_ETH_RX_BLOCK_PRIO_TAGGED |
+ MLX4_VLAN_CTRL_ETH_RX_BLOCK_UNTAGGED |
+ MLX4_VLAN_CTRL_ETH_RX_BLOCK_TAGGED;
+ else if (!work->vlan_id)
vlan_control = MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
MLX4_VLAN_CTRL_ETH_RX_BLOCK_TAGGED;
else
diff --git a/include/linux/mlx4/qp.h b/include/linux/mlx4/qp.h
index f43e32a..262deac 100644
--- a/include/linux/mlx4/qp.h
+++ b/include/linux/mlx4/qp.h
@@ -152,6 +152,8 @@ enum { /* fl */
};
enum { /* vlan_control */
MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED = 1 << 6,
+ MLX4_VLAN_CTRL_ETH_TX_BLOCK_PRIO_TAGGED = 1 << 5, /* 802.1p priority tag */
+ MLX4_VLAN_CTRL_ETH_TX_BLOCK_UNTAGGED = 1 << 4,
MLX4_VLAN_CTRL_ETH_RX_BLOCK_TAGGED = 1 << 2,
MLX4_VLAN_CTRL_ETH_RX_BLOCK_PRIO_TAGGED = 1 << 1, /* 802.1p priority tag */
MLX4_VLAN_CTRL_ETH_RX_BLOCK_UNTAGGED = 1 << 0
--
1.7.1
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH net-next 0/2] net/mlx4: Add dynamic policy changes to the SRIOV eSwitch support
2013-06-27 16:05 [PATCH net-next 0/2] net/mlx4: Add dynamic policy changes to the SRIOV eSwitch support Or Gerlitz
2013-06-27 16:05 ` [PATCH net-next 1/2] net/mlx4_core: Dynamic VST to VST vlan/qos changes Or Gerlitz
2013-06-27 16:05 ` [PATCH net-next 2/2] net/mlx4_core: Add HW enforcement to VF link state Or Gerlitz
@ 2013-07-01 20:12 ` David Miller
2013-07-02 4:40 ` Or Gerlitz
2 siblings, 1 reply; 5+ messages in thread
From: David Miller @ 2013-07-01 20:12 UTC (permalink / raw)
To: ogerlitz; +Cc: netdev, jackm, ronye
From: Or Gerlitz <ogerlitz@mellanox.com>
Date: Thu, 27 Jun 2013 19:05:20 +0300
> This patch set from Jack and Rony uses the UPDATE_QP new firmware command
> to allow for dynamic policy changes of PF controlled VF vlan/qos and link-state.
>
> Done against net-next commit 8599b52e14 "bonding: add an option to fail when any
> of arp_ip_target is inaccessible"
Applied, but in patch #2:
+ /* even if the UPDATE_QP command isn't supported, we still want
+ * to set this VF link according to the admin directive
+ */
Those last two lines start with a space, then TAB(s), which I had to
fix up. GIT even warns about this when I apply your patches.
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH net-next 0/2] net/mlx4: Add dynamic policy changes to the SRIOV eSwitch support
2013-07-01 20:12 ` [PATCH net-next 0/2] net/mlx4: Add dynamic policy changes to the SRIOV eSwitch support David Miller
@ 2013-07-02 4:40 ` Or Gerlitz
0 siblings, 0 replies; 5+ messages in thread
From: Or Gerlitz @ 2013-07-02 4:40 UTC (permalink / raw)
To: David Miller; +Cc: ogerlitz, netdev, jackm, ronye
On Mon, Jul 1, 2013 at 11:12 PM, David Miller <davem@davemloft.net> wrote:
>
> From: Or Gerlitz <ogerlitz@mellanox.com>
> Date: Thu, 27 Jun 2013 19:05:20 +0300
>
> > This patch set from Jack and Rony uses the UPDATE_QP new firmware
> > command
> > to allow for dynamic policy changes of PF controlled VF vlan/qos and
> > link-state.
> >
> > Done against net-next commit 8599b52e14 "bonding: add an option to fail
> > when any
> > of arp_ip_target is inaccessible"
>
> Applied, but in patch #2:
>
> + /* even if the UPDATE_QP command isn't supported, we still want
> + * to set this VF link according to the admin directive
> + */
>
> Those last two lines start with a space, then TAB(s), which I had to
> fix up. GIT even warns about this when I apply your patches.
thanks, last time similar thing happened I said that next time I will
double check, so I probably need to move into triple checking... I
will do little investigation later today to see how come there were no
complaints on my system.
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2013-07-02 4:40 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2013-06-27 16:05 [PATCH net-next 0/2] net/mlx4: Add dynamic policy changes to the SRIOV eSwitch support Or Gerlitz
2013-06-27 16:05 ` [PATCH net-next 1/2] net/mlx4_core: Dynamic VST to VST vlan/qos changes Or Gerlitz
2013-06-27 16:05 ` [PATCH net-next 2/2] net/mlx4_core: Add HW enforcement to VF link state Or Gerlitz
2013-07-01 20:12 ` [PATCH net-next 0/2] net/mlx4: Add dynamic policy changes to the SRIOV eSwitch support David Miller
2013-07-02 4:40 ` Or Gerlitz
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.