From 3609c4fb4d07d4285e96187598f54cb21e9e9b08 Mon Sep 17 00:00:00 2001
From: Shesha Sreenivasamurthy <shesha@cisco.com>
Date: Wed, 2 Sep 2015 08:57:24 -0700
Subject: [PATCH 2/9] mbuf: rearrange rte_mbuf metadata to suit vpp

Offload structure in the second cache line, next pointer in the
first cache line. Issue reported to Intel.
---
 .../linuxapp/eal/include/exec-env/rte_kni_common.h | 10 +++++++--
 lib/librte_mbuf/rte_mbuf.h                         | 25 ++++++++++++++--------
 2 files changed, 24 insertions(+), 11 deletions(-)

diff --git a/lib/librte_eal/linuxapp/eal/include/exec-env/rte_kni_common.h b/lib/librte_eal/linuxapp/eal/include/exec-env/rte_kni_common.h
index e9f38bd..d327f71 100644
--- a/lib/librte_eal/linuxapp/eal/include/exec-env/rte_kni_common.h
+++ b/lib/librte_eal/linuxapp/eal/include/exec-env/rte_kni_common.h
@@ -111,6 +111,10 @@ struct rte_kni_fifo {
  * The kernel image of the rte_mbuf struct, with only the relevant fields.
  * Padding is necessary to assure the offsets of these fields
  */
+/*
+ * offload in the second cache line, next in the first. Better for vpp
+ * at least as of right now.
+ */
 struct rte_kni_mbuf {
 	void *buf_addr __attribute__((__aligned__(RTE_CACHE_LINE_SIZE)));
 	char pad0[10];
@@ -121,16 +125,18 @@ struct rte_kni_mbuf {
 	char pad2[4];
 	uint32_t pkt_len;       /**< Total pkt len: sum of all segment data_len. */
 	uint16_t data_len;      /**< Amount of data in segment buffer. */
+	char pad3[2];
 #else
 	char pad2[2];
 	uint16_t data_len;      /**< Amount of data in segment buffer. */
 	uint32_t pkt_len;       /**< Total pkt len: sum of all segment data_len. */
+	char pad3[4];
 #endif
+	void *next;
 
 	/* fields on second cache line */
-	char pad3[8] __attribute__((__aligned__(RTE_CACHE_LINE_SIZE)));
+	char pad4[12] __attribute__((__aligned__(RTE_CACHE_LINE_SIZE)));
 	void *pool;
-	void *next;
 };
 
 /*
diff --git a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h
index 8c2db1b..61cbbd7 100644
--- a/lib/librte_mbuf/rte_mbuf.h
+++ b/lib/librte_mbuf/rte_mbuf.h
@@ -743,6 +743,12 @@ typedef uint64_t MARKER64[0]; /**< marker that allows us to overwrite 8 bytes
 /**
  * The generic rte_mbuf, containing a packet mbuf.
  */
+/*
+ * offload in the second cache line, next in the first. Better for vpp 
+ * at least as of right now.
+ * If you change this structure, you must change the user-mode
+ * version in rte_mbuf.h
+ */
 struct rte_mbuf {
 	MARKER cacheline0;
 
@@ -809,6 +815,16 @@ struct rte_mbuf {
 	uint16_t vlan_tci;        /**< VLAN Tag Control Identifier (CPU order) */
 	uint16_t vlan_tci_outer;  /**< Outer VLAN Tag Control Identifier (CPU order) */
 #endif /* RTE_NEXT_ABI */
+	struct rte_mbuf *next;    /**< Next segment of scattered packet. */
+
+	uint32_t seqn; /**< Sequence number. See also rte_reorder_insert() */
+#ifdef RTE_NEXT_ABI
+	uint16_t vlan_tci_outer;  /**< Outer VLAN Tag Control Identifier (CPU order) */
+#endif /* RTE_NEXT_ABI */
+
+	/* second cache line - fields only used in slow path or on TX */
+	MARKER cacheline1 __rte_cache_aligned;
+
 	union {
 		uint32_t rss;     /**< RSS hash result if RSS enabled */
 		struct {
@@ -828,21 +844,12 @@ struct rte_mbuf {
 		uint32_t usr;	  /**< User defined tags. See rte_distributor_process() */
 	} hash;                   /**< hash information */
 
-	uint32_t seqn; /**< Sequence number. See also rte_reorder_insert() */
-#ifdef RTE_NEXT_ABI
-	uint16_t vlan_tci_outer;  /**< Outer VLAN Tag Control Identifier (CPU order) */
-#endif /* RTE_NEXT_ABI */
-
-	/* second cache line - fields only used in slow path or on TX */
-	MARKER cacheline1 __rte_cache_aligned;
-
 	union {
 		void *userdata;   /**< Can be used for external metadata */
 		uint64_t udata64; /**< Allow 8-byte userdata on 32-bit */
 	};
 
 	struct rte_mempool *pool; /**< Pool from which mbuf was allocated. */
-	struct rte_mbuf *next;    /**< Next segment of scattered packet. */
 
 	/* fields to support TX offloads */
 	union {
-- 
2.5.0