bond: performance harvesting

- hash is great. But it is a bit too slow for the DP. Use direct array indexing
to quickly retrieve the slave interface.
- the algorithm used by flow hash is great. But it is a bit too slow for the DP.
Use l2_hash_hash() extracted from lb_hash.h which ECMP is using. It makes use
of intrinsic crc32 instruction set.
- shortcut modulo arithmetic when the operand is 2**x (where x up to 4) to
avoid division instruction.
- special case for link count == 1 in bond_tx_fn()
- use clib_mem_unaligned to access data for the packet to avoid alignment error
- Fix some typos for packet tracing.

Change-Id: I8eae3ad497061c5473aa675ba894ee0211120d25
Signed-off-by: Steven <sluong@cisco.com>

1 files changed, 1 insertions, 21 deletions

diff --git a/src/plugins/lb/lbhash.h b/src/plugins/lb/lbhash.h
index 10d3beab13b..5d2ff24f044 100644
--- a/src/plugins/lb/lbhash.h
+++ b/src/plugins/lb/lbhash.h
@@ -30,6 +30,7 @@
 #define LB_PLUGIN_LB_LBHASH_H_
 
 #include <vnet/vnet.h>
+#include <vppinfra/lb_hash_hash.h>
 
 #if defined (__SSE4_2__)
 #include <immintrin.h>
@@ -101,27 +102,6 @@ void lb_hash_free(lb_hash_t *h)
   vec_free(mem);
 }
 
-#if defined(clib_crc32c_uses_intrinsics) && !defined (__i386__)
-static_always_inline
-u32 lb_hash_hash(u64 k0, u64 k1, u64 k2, u64 k3, u64 k4)
-{
-  u64 val = 0;
-  val = crc32_u64(val, k0);
-  val = crc32_u64(val, k1);
-  val = crc32_u64(val, k2);
-  val = crc32_u64(val, k3);
-  val = crc32_u64(val, k4);
-  return (u32) val;
-}
-#else
-static_always_inline
-u32 lb_hash_hash(u64 k0, u64 k1, u64 k2, u64 k3, u64 k4)
-{
-  u64 tmp = k0 ^ k1 ^ k2 ^ k3 ^ k4;
-  return (u32)clib_xxhash (tmp);
-}
-#endif
-
 static_always_inline
 void lb_hash_prefetch_bucket(lb_hash_t *ht, u32 hash)
 {