Track number of ethernet vlans in a frame

Adds flags to the packet buffer to track the number of VLANs in
the current Ethernet frame. We use two bits to signify 0, 1 or
2 VLANs. The value 3 signififies an unknown quantity of VLANs,
which includes "three or more" which is not widely supported.
We place the bits in the vlib_buffer section; that is not the
opaque section, so that all subordinate nodes can use it.

For background, see the discussion thread at
https://lists.fd.io/pipermail/vpp-dev/2016-March/000354.html

The helper macro ethernet_buffer_header_size(buffer) uses
these bits stored in "buffer" to calculate the Ethernet header
size.

The macro ethernet_buffer_set_vlan_count(buffer, count) sets the
appropriate bit values based on the number in "count".

By current frame we are referring to the case where a packet
that arrives from the wire is carrying an encapsulated Ethernet
packet. Once decapsulated that inner packet becomes the current
frame.

There are two places where this value is set; For most Ethernet
frames this will be in the "ethernet-input" node when that node
parses the Ethernet header. The second place is whenever
vnet_update_l2_len() is used to update the layer 2 opaque data.
Typically this function is used by nodes just before they send
a packet into l2-input.

These bits are zeroed in vlib_buffer_init_for_free_list()
meaning that wherever the buffer comes from they have a reasonable
value (eg, if ip4/ip6 generates the packet.)

Primarily this VLAN counter is used by nodes below "ethernet-
input" and "l2-input" to determine where the start of the
current Ethernet header is. There is opaque data set by
"ethernet-input" storing the offset of the current Ethernet
header but, since this is opaque, it's not usable by downstream
nodes. Previously several nodes have made assumptions regarding
the location of the Ethernet header, including that it is always
at the start of the packet buffer (incorrect when we have
encapsulated packets) or that it is exactly
sizeof(ethernet_header_t) away (incorrect when we have VLAN tags.)

One notable case where this functionality is required is in
ip6_neighbor when it generates a response to a received neighbor
soliciation request; it reuses the incoming Ethernet header
in-situ and thus needs to reliably know where that header begins.

Also, at the suggestion of Dave Barach, this patch removes
definition of HGSHM bits in the buffer flags since they are
unused and unlikely to ever be.

Change-Id: I00e4b9ced5ef814a776020c395d1774aba6185b3
Signed-off-by: Chris Luke <chrisy@flirble.org>

9 files changed, 84 insertions, 9 deletions

diff --git a/vlib/vlib/buffer.h b/vlib/vlib/buffer.h
index 07ed85d8c31..e11085b72a0 100644
--- a/vlib/vlib/buffer.h
+++ b/vlib/vlib/buffer.h
@@ -93,8 +93,7 @@ typedef struct {
 #define VLIB_BUFFER_NEXT_PRESENT (1 << VLIB_BUFFER_LOG2_NEXT_PRESENT) 
 #define VLIB_BUFFER_IS_RECYCLED (1 << 2) 
 #define VLIB_BUFFER_TOTAL_LENGTH_VALID (1 << 3)
-#define VLIB_BUFFER_HGSHM_USER_INDEX_VALID (1 << 4) 
-#define VLIB_BUFFER_REPL_FAIL (1 << 5) 
+#define VLIB_BUFFER_REPL_FAIL (1 << 4) 
 
   /* User defined buffer flags. */
 #define LOG2_VLIB_BUFFER_FLAG_USER(n) (32 - (n))
diff --git a/vnet/vnet/buffer.h b/vnet/vnet/buffer.h
index f7fd8e8f872..7153f3541b1 100644
--- a/vnet/vnet/buffer.h
+++ b/vnet/vnet/buffer.h
@@ -49,6 +49,19 @@
 #define IP_BUFFER_L4_CHECKSUM_COMPUTED (1 << LOG2_IP_BUFFER_L4_CHECKSUM_COMPUTED)
 #define IP_BUFFER_L4_CHECKSUM_CORRECT  (1 << LOG2_IP_BUFFER_L4_CHECKSUM_CORRECT)
 
+/* VLAN header flags.
+ * These bits are zeroed in vlib_buffer_init_for_free_list()
+ * meaning wherever the buffer comes from they have a reasonable
+ * value (eg, if ip4/ip6 generates the packet.)
+ */
+#define LOG2_ETH_BUFFER_VLAN_2_DEEP LOG2_VLIB_BUFFER_FLAG_USER(3)
+#define LOG2_ETH_BUFFER_VLAN_1_DEEP LOG2_VLIB_BUFFER_FLAG_USER(4)
+#define ETH_BUFFER_VLAN_2_DEEP (1 << LOG2_ETH_BUFFER_VLAN_2_DEEP)
+#define ETH_BUFFER_VLAN_1_DEEP (1 << LOG2_ETH_BUFFER_VLAN_1_DEEP)
+#define ETH_BUFFER_VLAN_BITS (ETH_BUFFER_VLAN_1_DEEP | \
+                              ETH_BUFFER_VLAN_2_DEEP)
+
+
 #define foreach_buffer_opaque_union_subtype     \
 _(ethernet)                                     \
 _(ip)                                           \
diff --git a/vnet/vnet/classify/ip_classify.c b/vnet/vnet/classify/ip_classify.c
index 55c08e0ceeb..8152f6ce353 100644
--- a/vnet/vnet/classify/ip_classify.c
+++ b/vnet/vnet/classify/ip_classify.c
@@ -104,11 +104,13 @@ ip_classify_inline (vlib_main_t * vm,
         
       bi0 = from[0];
       b0 = vlib_get_buffer (vm, bi0);
-      h0 = b0->data;
+      h0 = (void *)vlib_buffer_get_current(b0) -
+                ethernet_buffer_header_size(b0);
 
       bi1 = from[1];
       b1 = vlib_get_buffer (vm, bi1);
-      h1 = b1->data;
+      h1 = (void *)vlib_buffer_get_current(b1) -
+                ethernet_buffer_header_size(b1);
         
       adj_index0 = vnet_buffer (b0)->ip.adj_index[VLIB_TX];
       adj0 = ip_get_adjacency (lm, adj_index0);
@@ -152,7 +154,8 @@ ip_classify_inline (vlib_main_t * vm,
 
       bi0 = from[0];
       b0 = vlib_get_buffer (vm, bi0);
-      h0 = b0->data;
+      h0 = (void *)vlib_buffer_get_current(b0) -
+                ethernet_buffer_header_size(b0);
         
       adj_index0 = vnet_buffer (b0)->ip.adj_index[VLIB_TX];
       adj0 = ip_get_adjacency (lm, adj_index0);
diff --git a/vnet/vnet/dhcpv6/proxy_node.c b/vnet/vnet/dhcpv6/proxy_node.c
index 01f5965c1b4..4dc746f6936 100644
--- a/vnet/vnet/dhcpv6/proxy_node.c
+++ b/vnet/vnet/dhcpv6/proxy_node.c
@@ -201,8 +201,7 @@ dhcpv6_proxy_to_server_input (vlib_main_t * vm,
 	  b0 = vlib_get_buffer (vm, bi0);
        
           h0 = vlib_buffer_get_current (b0);
-          e_h0 = (ethernet_header_t *)b0->data;
-          clib_memcpy(client_src_mac, e_h0->src_address, 6);
+
           /* Send to DHCPV6 server via the configured FIB */
           vnet_buffer(b0)->sw_if_index[VLIB_TX] =
               dpm->server_fib_index;
@@ -212,6 +211,9 @@ dhcpv6_proxy_to_server_input (vlib_main_t * vm,
            */
           u0 = (void *)h0 -(sizeof(*u0));
           ip0 = (void *)u0 -(sizeof(*ip0));
+          e_h0 = (void *)ip0 - ethernet_buffer_header_size(b0);
+
+          clib_memcpy(client_src_mac, e_h0->src_address, 6);
 
           switch (h0->u.msg_type) {
             case DHCPV6_MSG_SOLICIT:
diff --git a/vnet/vnet/ethernet/ethernet.h b/vnet/vnet/ethernet/ethernet.h
index 492aa759d00..6f16527c0e3 100644
--- a/vnet/vnet/ethernet/ethernet.h
+++ b/vnet/vnet/ethernet/ethernet.h
@@ -308,6 +308,50 @@ ethernet_buffer_get_header (vlib_buffer_t * b)
      + vnet_buffer (b)->ethernet.start_of_ethernet_header);
 }
 
+/** Returns the number of VLAN headers in the current Ethernet frame in the
+ * buffer. Returns 0, 1, 2 for the known header count. The value 3 indicates
+ * the number of headers is not known.
+ */
+#define ethernet_buffer_get_vlan_count(b) ( \
+    ((b)->flags & ETH_BUFFER_VLAN_BITS) >> LOG2_ETH_BUFFER_VLAN_1_DEEP \
+)
+
+/** Sets the number of VLAN headers in the current Ethernet frame in the
+ * buffer. Values 0, 1, 2 indicate  the header count. The value 3 indicates
+ * the number of headers is not known.
+ */
+#define ethernet_buffer_set_vlan_count(b, v) ( \
+    (b)->flags = ((b)->flags & ~ETH_BUFFER_VLAN_BITS) | \
+        (((v) << LOG2_ETH_BUFFER_VLAN_1_DEEP) & ETH_BUFFER_VLAN_BITS) \
+)
+
+/** Adjusts the vlan count by the delta in 'v' */
+#define ethernet_buffer_adjust_vlan_count(b, v) ( \
+  ethernet_buffer_set_vlan_count(b,  \
+      (word)ethernet_buffer_get_vlan_count(b) + (word)(v)) \
+)
+
+/** Adjusts the vlan count by the header size byte delta in 'v' */
+#define ethernet_buffer_adjust_vlan_count_by_bytes(b, v) ( \
+    (b)->flags = ((b)->flags & ~ETH_BUFFER_VLAN_BITS) | (( \
+        ((b)->flags & ETH_BUFFER_VLAN_BITS) + \
+        ((v) << (LOG2_ETH_BUFFER_VLAN_1_DEEP - 2)) \
+    ) & ETH_BUFFER_VLAN_BITS) \
+)
+
+/**
+ * Determine the size of the Ethernet headers of the current frame in
+ * the buffer. This uses the VLAN depth flags that are set by
+ * ethernet-input. Because these flags are stored in the vlib_buffer_t
+ * "flags" field this count is valid regardless of the node so long as it's
+ * checked downstream of ethernet-input; That is, the value is not stored in
+ * the opaque space.
+ */
+#define ethernet_buffer_header_size(b) ( \
+        ethernet_buffer_get_vlan_count((b)) * sizeof(ethernet_vlan_header_t) + \
+        sizeof(ethernet_header_t) \
+)
+
 ethernet_main_t * ethernet_get_main (vlib_main_t * vm);
 u32 ethernet_set_flags (vnet_main_t * vnm, u32 hw_if_index, u32 flags);
 void ethernet_sw_interface_set_l2_mode (vnet_main_t * vnm, u32 sw_if_index, u32 l2);
diff --git a/vnet/vnet/ethernet/node.c b/vnet/vnet/ethernet/node.c
index 1ec49c4b805..7b9924f5cde 100644
--- a/vnet/vnet/ethernet/node.c
+++ b/vnet/vnet/ethernet/node.c
@@ -98,6 +98,7 @@ parse_header (ethernet_input_variant_t variant,
               u16 * outer_id,
               u16 * inner_id,
               u32 * match_flags) {
+  u8 vlan_count;
 
   if (variant == ETHERNET_INPUT_VARIANT_ETHERNET 
       || variant == ETHERNET_INPUT_VARIANT_NOT_L2) {
@@ -129,6 +130,7 @@ parse_header (ethernet_input_variant_t variant,
   *inner_id = 0;
 
   *match_flags = SUBINT_CONFIG_VALID | SUBINT_CONFIG_MATCH_0_TAG;
+  vlan_count = 0;
 
   // check for vlan encaps
   if ((*type == ETHERNET_TYPE_VLAN) ||
@@ -150,6 +152,7 @@ parse_header (ethernet_input_variant_t variant,
     *type = clib_net_to_host_u16(h0->type);
 
     vlib_buffer_advance (b0, sizeof (h0[0]));
+    vlan_count = 1;
 
     if (*type == ETHERNET_TYPE_VLAN) {
       // Double tagged packet
@@ -164,13 +167,16 @@ parse_header (ethernet_input_variant_t variant,
       *type = clib_net_to_host_u16(h0->type);
 
       vlib_buffer_advance (b0, sizeof (h0[0]));
+      vlan_count = 2;
 
       if (*type == ETHERNET_TYPE_VLAN) {
         // More than double tagged packet
         *match_flags = SUBINT_CONFIG_VALID | SUBINT_CONFIG_MATCH_3_TAG;
+        vlan_count = 3; // "unknown" number, aka, 3-or-more
       }
     }
   }
+  ethernet_buffer_set_vlan_count(b0, vlan_count);
 }
 
 // Determine the subinterface for this packet, given the result of the
@@ -230,7 +236,7 @@ determine_next_node (ethernet_main_t * em,
     *next0 = em->l2_next;
     // record the L2 len and reset the buffer so the L2 header is preserved
     vnet_buffer(b0)->l2.l2_len = b0->current_data;
-    vlib_buffer_advance (b0, -(b0->current_data));
+    vlib_buffer_advance(b0, - ethernet_buffer_header_size(b0));
 
   // check for common IP/MPLS ethertypes
   } else if (type0 == ETHERNET_TYPE_IP4) {
diff --git a/vnet/vnet/ip/ip6_neighbor.c b/vnet/vnet/ip/ip6_neighbor.c
index 329cc6d7d67..19eb5a8d54d 100644
--- a/vnet/vnet/ip/ip6_neighbor.c
+++ b/vnet/vnet/ip/ip6_neighbor.c
@@ -847,7 +847,7 @@ icmp6_neighbor_solicitation_or_advertisement (vlib_main_t * vm,
 
               /* Reuse current MAC header, copy SMAC to DMAC and 
                * interface MAC to SMAC */
-              vlib_buffer_reset (p0);
+              vlib_buffer_advance(p0, - ethernet_buffer_header_size(p0));
               eth0 = vlib_buffer_get_current(p0);
               clib_memcpy(eth0->dst_address, eth0->src_address, 6);
               clib_memcpy(eth0->src_address, eth_if0->address, 6);
diff --git a/vnet/vnet/l2/l2_input.h b/vnet/vnet/l2/l2_input.h
index 1e5c1f0abc8..50649e79061 100644
--- a/vnet/vnet/l2/l2_input.h
+++ b/vnet/vnet/l2/l2_input.h
@@ -227,6 +227,7 @@ vnet_update_l2_len (vlib_buffer_t * b)
 {
   ethernet_header_t * eth;
   u16 ethertype;
+  u8 vlan_count = 0;
 
   /* point at currrent l2 hdr */
   eth = vlib_buffer_get_current (b);
@@ -245,12 +246,15 @@ vnet_update_l2_len (vlib_buffer_t * b)
       (ethertype == ETHERNET_TYPE_VLAN_9200)) {    
     ethernet_vlan_header_t * vlan;
     vnet_buffer(b)->l2.l2_len += sizeof (*vlan);
+    vlan_count = 1;
     vlan = (void *) (eth+1);
     ethertype = clib_net_to_host_u16 (vlan->type);
     if (ethertype == ETHERNET_TYPE_VLAN) {
       vnet_buffer(b)->l2.l2_len += sizeof (*vlan);
+      vlan_count = 2;
     }
   }
+  ethernet_buffer_set_vlan_count(b, vlan_count);
 }
 
 /*
diff --git a/vnet/vnet/l2/l2_vtr.h b/vnet/vnet/l2/l2_vtr.h
index aef6c6d255e..ccd0920db52 100644
--- a/vnet/vnet/l2/l2_vtr.h
+++ b/vnet/vnet/l2/l2_vtr.h
@@ -96,6 +96,10 @@ l2_vtr_process (vlib_buffer_t * b0,
   // Update l2_len
   vnet_buffer(b0)->l2.l2_len += (word)config->push_bytes - (word)config->pop_bytes;
 
+  // Update vlan tag count
+  ethernet_buffer_adjust_vlan_count_by_bytes(b0,
+                (word)config->push_bytes - (word)config->pop_bytes);
+
   // Update packet len
   vlib_buffer_advance(b0, (word)config->pop_bytes - (word)config->push_bytes);