/* Called directly after IPoIB netdevice was created to initialize SW structs */
void mlx5i_init(struct mlx5_core_dev *mdev,
struct net_device *netdev,
const struct mlx5e_profile *profile,
void *ppriv)
{
struct mlx5e_priv *priv = mlx5i_epriv(netdev);
/* priv init */
priv->mdev = mdev;
priv->netdev = netdev;
priv->profile = profile;
priv->ppriv = ppriv;
priv->hard_mtu = MLX5_IB_GRH_BYTES + MLX5_IPOIB_HARD_LEN;
mutex_init(&priv->state_lock);
mlx5e_build_nic_params(mdev, &priv->channels.params, profile->max_nch(mdev));
mlx5i_build_nic_params(mdev, &priv->channels.params);
/* netdev init */
netdev->hw_features |= NETIF_F_SG;
netdev->hw_features |= NETIF_F_IP_CSUM;
netdev->hw_features |= NETIF_F_IPV6_CSUM;
netdev->hw_features |= NETIF_F_GRO;
netdev->hw_features |= NETIF_F_TSO;
netdev->hw_features |= NETIF_F_TSO6;
netdev->hw_features |= NETIF_F_RXCSUM;
netdev->hw_features |= NETIF_F_RXHASH;
netdev->netdev_ops = &mlx5i_netdev_ops;
netdev->ethtool_ops = &mlx5i_ethtool_ops;
}
static int mlx5i_change_mtu(struct net_device *netdev, int new_mtu)
{
struct mlx5e_priv *priv = mlx5i_epriv(netdev);
struct mlx5e_channels new_channels = {};
struct mlx5e_params *params;
int err = 0;
mutex_lock(&priv->state_lock);
params = &priv->channels.params;
if (!test_bit(MLX5E_STATE_OPENED, &priv->state)) {
params->sw_mtu = new_mtu;
netdev->mtu = params->sw_mtu;
goto out;
}
new_channels.params = *params;
new_channels.params.sw_mtu = new_mtu;
err = mlx5e_open_channels(priv, &new_channels);
if (err)
goto out;
mlx5e_switch_priv_channels(priv, &new_channels, NULL);
netdev->mtu = new_channels.params.sw_mtu;
out:
mutex_unlock(&priv->state_lock);
return err;
}
static int mlx5i_close(struct net_device *netdev)
{
struct mlx5e_priv *epriv = mlx5i_epriv(netdev);
struct mlx5i_priv *ipriv = epriv->ppriv;
struct mlx5_core_dev *mdev = epriv->mdev;
/* May already be CLOSED in case a previous configuration operation
* (e.g RX/TX queue size change) that involves close&open failed.
*/
mutex_lock(&epriv->state_lock);
if (!test_bit(MLX5E_STATE_OPENED, &epriv->state))
goto unlock;
clear_bit(MLX5E_STATE_OPENED, &epriv->state);
netif_carrier_off(epriv->netdev);
mlx5_fs_remove_rx_underlay_qpn(mdev, ipriv->qp.qpn);
mlx5i_uninit_underlay_qp(epriv);
mlx5e_deactivate_priv_channels(epriv);
mlx5e_close_channels(&epriv->channels);
unlock:
mutex_unlock(&epriv->state_lock);
return 0;
}
static int mlx5i_pkey_change_mtu(struct net_device *netdev, int new_mtu)
{
struct mlx5e_priv *priv = mlx5i_epriv(netdev);
mutex_lock(&priv->state_lock);
netdev->mtu = new_mtu;
mutex_unlock(&priv->state_lock);
return 0;
}
void mlx5i_dev_cleanup(struct net_device *dev)
{
struct mlx5e_priv *priv = mlx5i_epriv(dev);
struct mlx5i_priv *ipriv = priv->ppriv;
mlx5i_uninit_underlay_qp(priv);
/* Delete QPN to net-device mapping from HT */
mlx5i_pkey_del_qpn(dev, ipriv->qp.qpn);
}
static int mlx5i_xmit(struct net_device *dev, struct sk_buff *skb,
struct ib_ah *address, u32 dqpn)
{
struct mlx5e_priv *epriv = mlx5i_epriv(dev);
struct mlx5e_txqsq *sq = epriv->txq2sq[skb_get_queue_mapping(skb)];
struct mlx5_ib_ah *mah = to_mah(address);
struct mlx5i_priv *ipriv = epriv->ppriv;
return mlx5i_sq_xmit(sq, skb, &mah->av, dqpn, ipriv->qkey);
}
int mlx5i_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
{
struct mlx5e_priv *priv = mlx5i_epriv(dev);
switch (cmd) {
case SIOCSHWTSTAMP:
return mlx5e_hwstamp_set(priv, ifr);
case SIOCGHWTSTAMP:
return mlx5e_hwstamp_get(priv, ifr);
default:
return -EOPNOTSUPP;
}
}
static int mlx5i_pkey_open(struct net_device *netdev)
{
struct mlx5e_priv *epriv = mlx5i_epriv(netdev);
struct mlx5i_priv *ipriv = epriv->ppriv;
struct mlx5_core_dev *mdev = epriv->mdev;
int err;
mutex_lock(&epriv->state_lock);
set_bit(MLX5E_STATE_OPENED, &epriv->state);
err = mlx5i_init_underlay_qp(epriv);
if (err) {
mlx5_core_warn(mdev, "prepare child underlay qp state failed, %d\n", err);
goto err_release_lock;
}
err = mlx5_fs_add_rx_underlay_qpn(mdev, ipriv->qp.qpn);
if (err) {
mlx5_core_warn(mdev, "attach child underlay qp to ft failed, %d\n", err);
goto err_unint_underlay_qp;
}
err = mlx5e_create_tis(mdev, 0 /* tc */, ipriv->qp.qpn, &epriv->tisn[0]);
if (err) {
mlx5_core_warn(mdev, "create child tis failed, %d\n", err);
goto err_remove_rx_uderlay_qp;
}
err = mlx5e_open_channels(epriv, &epriv->channels);
if (err) {
mlx5_core_warn(mdev, "opening child channels failed, %d\n", err);
goto err_clear_state_opened_flag;
}
mlx5e_refresh_tirs(epriv, false);
mlx5e_activate_priv_channels(epriv);
mutex_unlock(&epriv->state_lock);
return 0;
err_clear_state_opened_flag:
mlx5e_destroy_tis(mdev, epriv->tisn[0]);
err_remove_rx_uderlay_qp:
mlx5_fs_remove_rx_underlay_qpn(mdev, ipriv->qp.qpn);
err_unint_underlay_qp:
mlx5i_uninit_underlay_qp(epriv);
err_release_lock:
clear_bit(MLX5E_STATE_OPENED, &epriv->state);
mutex_unlock(&epriv->state_lock);
return err;
}
int mlx5i_dev_init(struct net_device *dev)
{
struct mlx5e_priv *priv = mlx5i_epriv(dev);
struct mlx5i_priv *ipriv = priv->ppriv;
/* Set dev address using underlay QP */
dev->dev_addr[1] = (ipriv->qp.qpn >> 16) & 0xff;
dev->dev_addr[2] = (ipriv->qp.qpn >> 8) & 0xff;
dev->dev_addr[3] = (ipriv->qp.qpn) & 0xff;
/* Add QPN to net-device mapping to HT */
mlx5i_pkey_add_qpn(dev ,ipriv->qp.qpn);
return 0;
}
void mlx5_rdma_netdev_free(struct net_device *netdev)
{
struct mlx5e_priv *priv = mlx5i_epriv(netdev);
struct mlx5i_priv *ipriv = priv->ppriv;
const struct mlx5e_profile *profile = priv->profile;
mlx5e_detach_netdev(priv);
profile->cleanup(priv);
destroy_workqueue(priv->wq);
if (!ipriv->sub_interface) {
mlx5i_pkey_qpn_ht_cleanup(netdev);
mlx5e_destroy_mdev_resources(priv->mdev);
}
free_netdev(netdev);
}
static int mlx5i_detach_mcast(struct net_device *netdev, struct ib_device *hca,
union ib_gid *gid, u16 lid)
{
struct mlx5e_priv *epriv = mlx5i_epriv(netdev);
struct mlx5_core_dev *mdev = epriv->mdev;
struct mlx5i_priv *ipriv = epriv->ppriv;
int err;
mlx5_core_dbg(mdev, "detaching QPN 0x%x, MGID %pI6\n", ipriv->qp.qpn, gid->raw);
err = mlx5_core_detach_mcg(mdev, gid, ipriv->qp.qpn);
if (err)
mlx5_core_dbg(mdev, "failed detaching QPN 0x%x, MGID %pI6\n",
ipriv->qp.qpn, gid->raw);
return err;
}
/* Called directly after IPoIB netdevice was created to initialize SW structs */
static void mlx5i_pkey_init(struct mlx5_core_dev *mdev,
struct net_device *netdev,
const struct mlx5e_profile *profile,
void *ppriv)
{
struct mlx5e_priv *priv = mlx5i_epriv(netdev);
mlx5i_init(mdev, netdev, profile, ppriv);
/* Override parent ndo */
netdev->netdev_ops = &mlx5i_pkey_netdev_ops;
/* Set child limited ethtool support */
netdev->ethtool_ops = &mlx5i_pkey_ethtool_ops;
/* Use dummy rqs */
priv->channels.params.log_rq_size = MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE;
}
static int mlx5i_open(struct net_device *netdev)
{
struct mlx5e_priv *epriv = mlx5i_epriv(netdev);
struct mlx5i_priv *ipriv = epriv->ppriv;
struct mlx5_core_dev *mdev = epriv->mdev;
int err;
mutex_lock(&epriv->state_lock);
set_bit(MLX5E_STATE_OPENED, &epriv->state);
err = mlx5i_init_underlay_qp(epriv);
if (err) {
mlx5_core_warn(mdev, "prepare underlay qp state failed, %d\n", err);
goto err_clear_state_opened_flag;
}
err = mlx5_fs_add_rx_underlay_qpn(mdev, ipriv->qp.qpn);
if (err) {
mlx5_core_warn(mdev, "attach underlay qp to ft failed, %d\n", err);
goto err_reset_qp;
}
err = mlx5e_open_channels(epriv, &epriv->channels);
if (err)
goto err_remove_fs_underlay_qp;
mlx5e_refresh_tirs(epriv, false);
mlx5e_activate_priv_channels(epriv);
mlx5e_timestamp_set(epriv);
mutex_unlock(&epriv->state_lock);
return 0;
err_remove_fs_underlay_qp:
mlx5_fs_remove_rx_underlay_qpn(mdev, ipriv->qp.qpn);
err_reset_qp:
mlx5i_uninit_underlay_qp(epriv);
err_clear_state_opened_flag:
clear_bit(MLX5E_STATE_OPENED, &epriv->state);
mutex_unlock(&epriv->state_lock);
return err;
}
int mlx5i_pkey_del_qpn(struct net_device *netdev, u32 qpn)
{
struct mlx5e_priv *epriv = mlx5i_epriv(netdev);
struct mlx5i_priv *ipriv = epriv->ppriv;
struct mlx5i_pkey_qpn_ht *ht = ipriv->qpn_htbl;
struct qpn_to_netdev *node;
node = mlx5i_find_qpn_to_netdev_node(ht->buckets, qpn);
if (!node) {
mlx5_core_warn(epriv->mdev, "QPN to netdev delete from HT failed\n");
return -EINVAL;
}
spin_lock_bh(&ht->ht_lock);
hlist_del_init(&node->hlist);
spin_unlock_bh(&ht->ht_lock);
kfree(node);
return 0;
}
/* IPoIB RDMA netdev callbacks */
static int mlx5i_attach_mcast(struct net_device *netdev, struct ib_device *hca,
union ib_gid *gid, u16 lid, int set_qkey,
u32 qkey)
{
struct mlx5e_priv *epriv = mlx5i_epriv(netdev);
struct mlx5_core_dev *mdev = epriv->mdev;
struct mlx5i_priv *ipriv = epriv->ppriv;
int err;
mlx5_core_dbg(mdev, "attaching QPN 0x%x, MGID %pI6\n", ipriv->qp.qpn, gid->raw);
err = mlx5_core_attach_mcg(mdev, gid, ipriv->qp.qpn);
if (err)
mlx5_core_warn(mdev, "failed attaching QPN 0x%x, MGID %pI6\n",
ipriv->qp.qpn, gid->raw);
if (set_qkey) {
mlx5_core_dbg(mdev, "%s setting qkey 0x%x\n",
netdev->name, qkey);
ipriv->qkey = qkey;
}
return err;
}
static int mlx5i_pkey_close(struct net_device *netdev)
{
struct mlx5e_priv *priv = mlx5i_epriv(netdev);
struct mlx5i_priv *ipriv = priv->ppriv;
struct mlx5_core_dev *mdev = priv->mdev;
mutex_lock(&priv->state_lock);
if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
goto unlock;
clear_bit(MLX5E_STATE_OPENED, &priv->state);
netif_carrier_off(priv->netdev);
mlx5_fs_remove_rx_underlay_qpn(mdev, ipriv->qp.qpn);
mlx5i_uninit_underlay_qp(priv);
mlx5e_deactivate_priv_channels(priv);
mlx5e_close_channels(&priv->channels);
mlx5e_destroy_tis(mdev, priv->tisn[0]);
unlock:
mutex_unlock(&priv->state_lock);
return 0;
}
/* Child NDOs */
static int mlx5i_pkey_dev_init(struct net_device *dev)
{
struct mlx5e_priv *priv = mlx5i_epriv(dev);
struct mlx5i_priv *ipriv, *parent_ipriv;
struct net_device *parent_dev;
int parent_ifindex;
ipriv = priv->ppriv;
/* Get QPN to netdevice hash table from parent */
parent_ifindex = dev->netdev_ops->ndo_get_iflink(dev);
parent_dev = dev_get_by_index(dev_net(dev), parent_ifindex);
if (!parent_dev) {
mlx5_core_warn(priv->mdev, "failed to get parent device\n");
return -EINVAL;
}
parent_ipriv = netdev_priv(parent_dev);
ipriv->qpn_htbl = parent_ipriv->qpn_htbl;
dev_put(parent_dev);
return mlx5i_dev_init(dev);
}
/* Called directly after IPoIB netdevice was created to initialize SW structs */
void mlx5i_init(struct mlx5_core_dev *mdev,
struct net_device *netdev,
const struct mlx5e_profile *profile,
void *ppriv)
{
struct mlx5e_priv *priv = mlx5i_epriv(netdev);
u16 max_mtu;
/* priv init */
priv->mdev = mdev;
priv->netdev = netdev;
priv->profile = profile;
priv->ppriv = ppriv;
mutex_init(&priv->state_lock);
mlx5_query_port_max_mtu(mdev, &max_mtu, 1);
netdev->mtu = max_mtu;
mlx5e_build_nic_params(mdev, &priv->channels.params,
profile->max_nch(mdev), netdev->mtu);
mlx5i_build_nic_params(mdev, &priv->channels.params);
mlx5e_timestamp_init(priv);
/* netdev init */
netdev->hw_features |= NETIF_F_SG;
netdev->hw_features |= NETIF_F_IP_CSUM;
netdev->hw_features |= NETIF_F_IPV6_CSUM;
netdev->hw_features |= NETIF_F_GRO;
netdev->hw_features |= NETIF_F_TSO;
netdev->hw_features |= NETIF_F_TSO6;
netdev->hw_features |= NETIF_F_RXCSUM;
netdev->hw_features |= NETIF_F_RXHASH;
netdev->netdev_ops = &mlx5i_netdev_ops;
netdev->ethtool_ops = &mlx5i_ethtool_ops;
}
struct net_device *mlx5_rdma_netdev_alloc(struct mlx5_core_dev *mdev,
struct ib_device *ibdev,
const char *name,
void (*setup)(struct net_device *))
{
const struct mlx5e_profile *profile;
struct net_device *netdev;
struct mlx5i_priv *ipriv;
struct mlx5e_priv *epriv;
struct rdma_netdev *rn;
bool sub_interface;
int nch;
int err;
if (mlx5i_check_required_hca_cap(mdev)) {
mlx5_core_warn(mdev, "Accelerated mode is not supported\n");
return ERR_PTR(-EOPNOTSUPP);
}
/* TODO: Need to find a better way to check if child device*/
sub_interface = (mdev->mlx5e_res.pdn != 0);
if (sub_interface)
profile = mlx5i_pkey_get_profile();
else
profile = &mlx5i_nic_profile;
nch = profile->max_nch(mdev);
netdev = alloc_netdev_mqs(sizeof(struct mlx5i_priv) + sizeof(struct mlx5e_priv),
name, NET_NAME_UNKNOWN,
setup,
nch * MLX5E_MAX_NUM_TC,
nch);
if (!netdev) {
mlx5_core_warn(mdev, "alloc_netdev_mqs failed\n");
return NULL;
}
ipriv = netdev_priv(netdev);
epriv = mlx5i_epriv(netdev);
epriv->wq = create_singlethread_workqueue("mlx5i");
if (!epriv->wq)
goto err_free_netdev;
ipriv->sub_interface = sub_interface;
if (!ipriv->sub_interface) {
err = mlx5i_pkey_qpn_ht_init(netdev);
if (err) {
mlx5_core_warn(mdev, "allocate qpn_to_netdev ht failed\n");
goto destroy_wq;
}
/* This should only be called once per mdev */
err = mlx5e_create_mdev_resources(mdev);
if (err)
goto destroy_ht;
}
profile->init(mdev, netdev, profile, ipriv);
mlx5e_attach_netdev(epriv);
netif_carrier_off(netdev);
/* set rdma_netdev func pointers */
rn = &ipriv->rn;
rn->hca = ibdev;
rn->send = mlx5i_xmit;
rn->attach_mcast = mlx5i_attach_mcast;
rn->detach_mcast = mlx5i_detach_mcast;
rn->set_id = mlx5i_set_pkey_index;
return netdev;
destroy_ht:
mlx5i_pkey_qpn_ht_cleanup(netdev);
destroy_wq:
destroy_workqueue(epriv->wq);
err_free_netdev:
free_netdev(netdev);
return NULL;
}
