bnxt_en: Improve ntuple filters by checking destination MAC address.
authorMichael Chan <michael.chan@broadcom.com>
Mon, 25 Jul 2016 16:33:35 +0000 (12:33 -0400)
committerDavid S. Miller <davem@davemloft.net>
Tue, 26 Jul 2016 04:43:31 +0000 (21:43 -0700)
Include the destination MAC address in the ntuple filter structure.  The
current code assumes that the destination MAC address is always the MAC
address of the NIC.  This may not be true if there are macvlans, for
example.  Add destination MAC address checking and configure the filter
correctly using the correct index for the destination MAC address.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/ethernet/broadcom/bnxt/bnxt.c
drivers/net/ethernet/broadcom/bnxt/bnxt.h

index 8a0165b0b684b19724c84592243c53c9a3fd71fc..7de7d7ac7c3a6584c6645d57832e521996700b8a 100644 (file)
@@ -3240,7 +3240,7 @@ static int bnxt_hwrm_cfa_ntuple_filter_alloc(struct bnxt *bp,
        struct bnxt_vnic_info *vnic = &bp->vnic_info[fltr->rxq + 1];
 
        bnxt_hwrm_cmd_hdr_init(bp, &req, HWRM_CFA_NTUPLE_FILTER_ALLOC, -1, -1);
-       req.l2_filter_id = bp->vnic_info[0].fw_l2_filter_id[0];
+       req.l2_filter_id = bp->vnic_info[0].fw_l2_filter_id[fltr->l2_fltr_idx];
 
        req.enables = cpu_to_le32(BNXT_NTP_FLTR_FLAGS);
 
@@ -6299,7 +6299,8 @@ static bool bnxt_fltr_match(struct bnxt_ntuple_filter *f1,
            keys1->ports.ports == keys2->ports.ports &&
            keys1->basic.ip_proto == keys2->basic.ip_proto &&
            keys1->basic.n_proto == keys2->basic.n_proto &&
-           ether_addr_equal(f1->src_mac_addr, f2->src_mac_addr))
+           ether_addr_equal(f1->src_mac_addr, f2->src_mac_addr) &&
+           ether_addr_equal(f1->dst_mac_addr, f2->dst_mac_addr))
                return true;
 
        return false;
@@ -6312,12 +6313,28 @@ static int bnxt_rx_flow_steer(struct net_device *dev, const struct sk_buff *skb,
        struct bnxt_ntuple_filter *fltr, *new_fltr;
        struct flow_keys *fkeys;
        struct ethhdr *eth = (struct ethhdr *)skb_mac_header(skb);
-       int rc = 0, idx, bit_id;
+       int rc = 0, idx, bit_id, l2_idx = 0;
        struct hlist_head *head;
 
        if (skb->encapsulation)
                return -EPROTONOSUPPORT;
 
+       if (!ether_addr_equal(dev->dev_addr, eth->h_dest)) {
+               struct bnxt_vnic_info *vnic = &bp->vnic_info[0];
+               int off = 0, j;
+
+               netif_addr_lock_bh(dev);
+               for (j = 0; j < vnic->uc_filter_count; j++, off += ETH_ALEN) {
+                       if (ether_addr_equal(eth->h_dest,
+                                            vnic->uc_list + off)) {
+                               l2_idx = j + 1;
+                               break;
+                       }
+               }
+               netif_addr_unlock_bh(dev);
+               if (!l2_idx)
+                       return -EINVAL;
+       }
        new_fltr = kzalloc(sizeof(*new_fltr), GFP_ATOMIC);
        if (!new_fltr)
                return -ENOMEM;
@@ -6335,6 +6352,7 @@ static int bnxt_rx_flow_steer(struct net_device *dev, const struct sk_buff *skb,
                goto err_free;
        }
 
+       memcpy(new_fltr->dst_mac_addr, eth->h_dest, ETH_ALEN);
        memcpy(new_fltr->src_mac_addr, eth->h_source, ETH_ALEN);
 
        idx = skb_get_hash_raw(skb) & BNXT_NTP_FLTR_HASH_MASK;
@@ -6360,6 +6378,7 @@ static int bnxt_rx_flow_steer(struct net_device *dev, const struct sk_buff *skb,
 
        new_fltr->sw_id = (u16)bit_id;
        new_fltr->flow_id = flow_id;
+       new_fltr->l2_fltr_idx = l2_idx;
        new_fltr->rxq = rxq_index;
        hlist_add_head_rcu(&new_fltr->hash, head);
        bp->ntp_fltr_count++;
index 5307a2ee6a05eae25e002e310ca5618ddc6cceda..23e04a6142fbec06fa027bde6f94f7f5d9e49e83 100644 (file)
@@ -785,10 +785,12 @@ struct bnxt_pf_info {
 
 struct bnxt_ntuple_filter {
        struct hlist_node       hash;
+       u8                      dst_mac_addr[ETH_ALEN];
        u8                      src_mac_addr[ETH_ALEN];
        struct flow_keys        fkeys;
        __le64                  filter_id;
        u16                     sw_id;
+       u8                      l2_fltr_idx;
        u16                     rxq;
        u32                     flow_id;
        unsigned long           state;