The default fallback function used by mlx5e select queue can return
any TX queues in range [0..dev->num_real_tx_queues).
The current implementation assumes that the fallback function returns
a number in the range [0.. number of channels). Actually
dev->num_real_tx_queues = (number of channels) * dev->num_tc;
which is more than the expected range if num_tc is configured and could
lead to crashes.
To fix this we test if num_tc is not configured we can safely return the
fallback suggestion, if not we will reciprocal_scale the fallback
result and normalize it to the desired range.
Fixes:
08fb1dacdd76 ('net/mlx5e: Support DCBNL IEEE ETS')
Signed-off-by: Rana Shahout <ranas@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Reported-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
netdev_set_num_tc(netdev, ntc);
+ /* Map netdev TCs to offset 0
+ * We have our own UP to TXQ mapping for QoS
+ */
for (tc = 0; tc < ntc; tc++)
- netdev_set_tc_queue(netdev, tc, nch, tc * nch);
+ netdev_set_tc_queue(netdev, tc, nch, 0);
}
int mlx5e_open_locked(struct net_device *netdev)
{
struct mlx5e_priv *priv = netdev_priv(dev);
int channel_ix = fallback(dev, skb);
- int up = (netdev_get_num_tc(dev) && skb_vlan_tag_present(skb)) ?
- skb->vlan_tci >> VLAN_PRIO_SHIFT : 0;
+ int up = 0;
+
+ if (!netdev_get_num_tc(dev))
+ return channel_ix;
+
+ if (skb_vlan_tag_present(skb))
+ up = skb->vlan_tci >> VLAN_PRIO_SHIFT;
+
+ /* channel_ix can be larger than num_channels since
+ * dev->num_real_tx_queues = num_channels * num_tc
+ */
+ if (channel_ix >= priv->params.num_channels)
+ channel_ix = reciprocal_scale(channel_ix,
+ priv->params.num_channels);
return priv->channeltc_to_txq_map[channel_ix][up];
}