[dpdk-dev] [PATCH v4 5/9] eal: add and use unaligned integer types

[Cyril Chemparathy <cchemparathy@ezchip.com>]

On machines that are strict on pointer alignment, current code breaks
on GCC's -Wcast-align checks on casts from narrower to wider types.
This patch introduces new unaligned_uint(16|32|64)_t types, which
correctly retain alignment in such cases.  Strict alignment
architectures will need to define CONFIG_RTE_ARCH_STRICT_ALIGN in
order to effect these new types.

Change-Id: I6f7c429fc14233c991287d065ecf2a09dbd65ebb
Signed-off-by: Cyril Chemparathy <cchemparathy@ezchip.com>
---
 app/test-pmd/flowgen.c                     |  4 ++--
 app/test-pmd/icmpecho.c                    |  2 +-
 app/test-pmd/txonly.c                      |  2 +-
 app/test/packet_burst_generator.c          |  4 ++--
 app/test/test_hash_functions.c             |  2 +-
 app/test/test_mbuf.c                       | 16 ++++++++--------
 config/common_bsdapp                       |  5 +++++
 config/common_linuxapp                     |  5 +++++
 drivers/net/bonding/rte_eth_bond_pmd.c     | 12 ++++++++----
 lib/librte_eal/common/include/rte_common.h | 10 ++++++++++
 lib/librte_ether/rte_ether.h               |  2 +-
 lib/librte_ip_frag/rte_ipv4_reassembly.c   |  4 ++--
 12 files changed, 46 insertions(+), 22 deletions(-)

diff --git a/app/test-pmd/flowgen.c b/app/test-pmd/flowgen.c
index 72016c9..174c003 100644
--- a/app/test-pmd/flowgen.c
+++ b/app/test-pmd/flowgen.c
@@ -101,7 +101,7 @@ tx_mbuf_alloc(struct rte_mempool *mp)
 
 
 static inline uint16_t
-ip_sum(const uint16_t *hdr, int hdr_len)
+ip_sum(const unaligned_uint16_t *hdr, int hdr_len)
 {
 	uint32_t sum = 0;
 
@@ -193,7 +193,7 @@ pkt_burst_flow_gen(struct fwd_stream *fs)
 							   next_flow);
 		ip_hdr->total_length	= RTE_CPU_TO_BE_16(pkt_size -
 							   sizeof(*eth_hdr));
-		ip_hdr->hdr_checksum	= ip_sum((uint16_t *)ip_hdr,
+		ip_hdr->hdr_checksum	= ip_sum((unaligned_uint16_t *)ip_hdr,
 						 sizeof(*ip_hdr));
 
 		/* Initialize UDP header. */
diff --git a/app/test-pmd/icmpecho.c b/app/test-pmd/icmpecho.c
index 29aef71..e510f9b 100644
--- a/app/test-pmd/icmpecho.c
+++ b/app/test-pmd/icmpecho.c
@@ -282,7 +282,7 @@ ipv4_hdr_cksum(struct ipv4_hdr *ip_h)
 	 * Compute the sum of successive 16-bit words of the IPv4 header,
 	 * skipping the checksum field of the header.
 	 */
-	v16_h = (uint16_t *) ip_h;
+	v16_h = (unaligned_uint16_t *) ip_h;
 	ip_cksum = v16_h[0] + v16_h[1] + v16_h[2] + v16_h[3] +
 		v16_h[4] + v16_h[6] + v16_h[7] + v16_h[8] + v16_h[9];
 
diff --git a/app/test-pmd/txonly.c b/app/test-pmd/txonly.c
index ca32c85..9e66552 100644
--- a/app/test-pmd/txonly.c
+++ b/app/test-pmd/txonly.c
@@ -167,7 +167,7 @@ setup_pkt_udp_ip_headers(struct ipv4_hdr *ip_hdr,
 	/*
 	 * Compute IP header checksum.
 	 */
-	ptr16 = (uint16_t*) ip_hdr;
+	ptr16 = (unaligned_uint16_t*) ip_hdr;
 	ip_cksum = 0;
 	ip_cksum += ptr16[0]; ip_cksum += ptr16[1];
 	ip_cksum += ptr16[2]; ip_cksum += ptr16[3];
diff --git a/app/test/packet_burst_generator.c b/app/test/packet_burst_generator.c
index b46eed7..06762c6 100644
--- a/app/test/packet_burst_generator.c
+++ b/app/test/packet_burst_generator.c
@@ -154,7 +154,7 @@ initialize_ipv4_header(struct ipv4_hdr *ip_hdr, uint32_t src_addr,
 		uint32_t dst_addr, uint16_t pkt_data_len)
 {
 	uint16_t pkt_len;
-	uint16_t *ptr16;
+	unaligned_uint16_t *ptr16;
 	uint32_t ip_cksum;
 
 	/*
@@ -175,7 +175,7 @@ initialize_ipv4_header(struct ipv4_hdr *ip_hdr, uint32_t src_addr,
 	/*
 	 * Compute IP header checksum.
 	 */
-	ptr16 = (uint16_t *)ip_hdr;
+	ptr16 = (unaligned_uint16_t *)ip_hdr;
 	ip_cksum = 0;
 	ip_cksum += ptr16[0]; ip_cksum += ptr16[1];
 	ip_cksum += ptr16[2]; ip_cksum += ptr16[3];
diff --git a/app/test/test_hash_functions.c b/app/test/test_hash_functions.c
index df7c909..8c7cf63 100644
--- a/app/test/test_hash_functions.c
+++ b/app/test/test_hash_functions.c
@@ -223,7 +223,7 @@ verify_jhash_32bits(void)
 				hash = rte_jhash(key, hashtest_key_lens[i],
 						hashtest_initvals[j]);
 				/* Divide key length by 4 in rte_jhash for 32 bits */
-				hash32 = rte_jhash_32b((const uint32_t *)key,
+				hash32 = rte_jhash_32b((const unaligned_uint32_t *)key,
 						hashtest_key_lens[i] >> 2,
 						hashtest_initvals[j]);
 				if (hash != hash32) {
diff --git a/app/test/test_mbuf.c b/app/test/test_mbuf.c
index 5e8a377..d9beb29 100644
--- a/app/test/test_mbuf.c
+++ b/app/test/test_mbuf.c
@@ -333,7 +333,7 @@ testclone_testupdate_testdetach(void)
 	struct rte_mbuf *m = NULL;
 	struct rte_mbuf *clone = NULL;
 	struct rte_mbuf *clone2 = NULL;
-	uint32_t *data;
+	unaligned_uint32_t *data;
 
 	/* alloc a mbuf */
 	m = rte_pktmbuf_alloc(pktmbuf_pool);
@@ -344,7 +344,7 @@ testclone_testupdate_testdetach(void)
 		GOTO_FAIL("Bad length");
 
 	rte_pktmbuf_append(m, sizeof(uint32_t));
-	data = rte_pktmbuf_mtod(m, uint32_t *);
+	data = rte_pktmbuf_mtod(m, unaligned_uint32_t *);
 	*data = MAGIC_DATA;
 
 	/* clone the allocated mbuf */
@@ -352,7 +352,7 @@ testclone_testupdate_testdetach(void)
 	if (clone == NULL)
 		GOTO_FAIL("cannot clone data\n");
 
-	data = rte_pktmbuf_mtod(clone, uint32_t *);
+	data = rte_pktmbuf_mtod(clone, unaligned_uint32_t *);
 	if (*data != MAGIC_DATA)
 		GOTO_FAIL("invalid data in clone\n");
 
@@ -369,18 +369,18 @@ testclone_testupdate_testdetach(void)
 		GOTO_FAIL("Next Pkt Null\n");
 
 	rte_pktmbuf_append(m->next, sizeof(uint32_t));
-	data = rte_pktmbuf_mtod(m->next, uint32_t *);
+	data = rte_pktmbuf_mtod(m->next, unaligned_uint32_t *);
 	*data = MAGIC_DATA;
 
 	clone = rte_pktmbuf_clone(m, pktmbuf_pool);
 	if (clone == NULL)
 		GOTO_FAIL("cannot clone data\n");
 
-	data = rte_pktmbuf_mtod(clone, uint32_t *);
+	data = rte_pktmbuf_mtod(clone, unaligned_uint32_t *);
 	if (*data != MAGIC_DATA)
 		GOTO_FAIL("invalid data in clone\n");
 
-	data = rte_pktmbuf_mtod(clone->next, uint32_t *);
+	data = rte_pktmbuf_mtod(clone->next, unaligned_uint32_t *);
 	if (*data != MAGIC_DATA)
 		GOTO_FAIL("invalid data in clone->next\n");
 
@@ -396,11 +396,11 @@ testclone_testupdate_testdetach(void)
 	if (clone2 == NULL)
 		GOTO_FAIL("cannot clone the clone\n");
 
-	data = rte_pktmbuf_mtod(clone2, uint32_t *);
+	data = rte_pktmbuf_mtod(clone2, unaligned_uint32_t *);
 	if (*data != MAGIC_DATA)
 		GOTO_FAIL("invalid data in clone2\n");
 
-	data = rte_pktmbuf_mtod(clone2->next, uint32_t *);
+	data = rte_pktmbuf_mtod(clone2->next, unaligned_uint32_t *);
 	if (*data != MAGIC_DATA)
 		GOTO_FAIL("invalid data in clone2->next\n");
 
diff --git a/config/common_bsdapp b/config/common_bsdapp
index 0b169c8..3af59d9 100644
--- a/config/common_bsdapp
+++ b/config/common_bsdapp
@@ -74,6 +74,11 @@ CONFIG_RTE_EXEC_ENV_BSDAPP=y
 CONFIG_RTE_FORCE_INTRINSICS=n
 
 #
+# Machine forces strict alignment constraints.
+#
+CONFIG_RTE_ARCH_STRICT_ALIGN=n
+
+#
 # Compile to share library
 #
 CONFIG_RTE_BUILD_SHARED_LIB=n
diff --git a/config/common_linuxapp b/config/common_linuxapp
index 5deb55a..a0b37d8 100644
--- a/config/common_linuxapp
+++ b/config/common_linuxapp
@@ -74,6 +74,11 @@ CONFIG_RTE_EXEC_ENV_LINUXAPP=y
 CONFIG_RTE_FORCE_INTRINSICS=n
 
 #
+# Machine forces strict alignment constraints.
+#
+CONFIG_RTE_ARCH_STRICT_ALIGN=n
+
+#
 # Compile to share library
 #
 CONFIG_RTE_BUILD_SHARED_LIB=n
diff --git a/drivers/net/bonding/rte_eth_bond_pmd.c b/drivers/net/bonding/rte_eth_bond_pmd.c
index 8bad2e1..5a2fbef 100644
--- a/drivers/net/bonding/rte_eth_bond_pmd.c
+++ b/drivers/net/bonding/rte_eth_bond_pmd.c
@@ -466,8 +466,10 @@ bond_ethdev_tx_burst_active_backup(void *queue,
 static inline uint16_t
 ether_hash(struct ether_hdr *eth_hdr)
 {
-	uint16_t *word_src_addr = (uint16_t *)eth_hdr->s_addr.addr_bytes;
-	uint16_t *word_dst_addr = (uint16_t *)eth_hdr->d_addr.addr_bytes;
+	unaligned_uint16_t *word_src_addr =
+		(unaligned_uint16_t *)eth_hdr->s_addr.addr_bytes;
+	unaligned_uint16_t *word_dst_addr =
+		(unaligned_uint16_t *)eth_hdr->d_addr.addr_bytes;
 
 	return (word_src_addr[0] ^ word_dst_addr[0]) ^
 			(word_src_addr[1] ^ word_dst_addr[1]) ^
@@ -483,8 +485,10 @@ ipv4_hash(struct ipv4_hdr *ipv4_hdr)
 static inline uint32_t
 ipv6_hash(struct ipv6_hdr *ipv6_hdr)
 {
-	uint32_t *word_src_addr = (uint32_t *)&(ipv6_hdr->src_addr[0]);
-	uint32_t *word_dst_addr = (uint32_t *)&(ipv6_hdr->dst_addr[0]);
+	unaligned_uint32_t *word_src_addr =
+		(unaligned_uint32_t *)&(ipv6_hdr->src_addr[0]);
+	unaligned_uint32_t *word_dst_addr =
+		(unaligned_uint32_t *)&(ipv6_hdr->dst_addr[0]);
 
 	return (word_src_addr[0] ^ word_dst_addr[0]) ^
 			(word_src_addr[1] ^ word_dst_addr[1]) ^
diff --git a/lib/librte_eal/common/include/rte_common.h b/lib/librte_eal/common/include/rte_common.h
index c6e076b..dcff889 100644
--- a/lib/librte_eal/common/include/rte_common.h
+++ b/lib/librte_eal/common/include/rte_common.h
@@ -59,6 +59,16 @@ extern "C" {
 #define asm __asm__
 #endif
 
+#ifdef RTE_ARCH_STRICT_ALIGN
+typedef uint64_t unaligned_uint64_t __attribute__ ((aligned(1)));
+typedef uint32_t unaligned_uint32_t __attribute__ ((aligned(1)));
+typedef uint16_t unaligned_uint16_t __attribute__ ((aligned(1)));
+#else
+typedef uint64_t unaligned_uint64_t;
+typedef uint32_t unaligned_uint32_t;
+typedef uint16_t unaligned_uint16_t;
+#endif
+
 /*********** Macros to eliminate unused variable warnings ********/
 
 /**
diff --git a/lib/librte_ether/rte_ether.h b/lib/librte_ether/rte_ether.h
index a14308e..07c17d7 100644
--- a/lib/librte_ether/rte_ether.h
+++ b/lib/librte_ether/rte_ether.h
@@ -175,7 +175,7 @@ static inline int is_multicast_ether_addr(const struct ether_addr *ea)
  */
 static inline int is_broadcast_ether_addr(const struct ether_addr *ea)
 {
-	const uint16_t *ea_words = (const uint16_t *)ea;
+	const unaligned_uint16_t *ea_words = (const unaligned_uint16_t *)ea;
 
 	return (ea_words[0] == 0xFFFF && ea_words[1] == 0xFFFF &&
 		ea_words[2] == 0xFFFF);
diff --git a/lib/librte_ip_frag/rte_ipv4_reassembly.c b/lib/librte_ip_frag/rte_ipv4_reassembly.c
index 841ac14..279be6c 100644
--- a/lib/librte_ip_frag/rte_ipv4_reassembly.c
+++ b/lib/librte_ip_frag/rte_ipv4_reassembly.c
@@ -120,7 +120,7 @@ rte_ipv4_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
 {
 	struct ip_frag_pkt *fp;
 	struct ip_frag_key key;
-	const uint64_t *psd;
+	const unaligned_uint64_t *psd;
 	uint16_t ip_len;
 	uint16_t flag_offset, ip_ofs, ip_flag;
 
@@ -128,7 +128,7 @@ rte_ipv4_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
 	ip_ofs = (uint16_t)(flag_offset & IPV4_HDR_OFFSET_MASK);
 	ip_flag = (uint16_t)(flag_offset & IPV4_HDR_MF_FLAG);
 
-	psd = (uint64_t *)&ip_hdr->src_addr;
+	psd = (unaligned_uint64_t *)&ip_hdr->src_addr;
 	/* use first 8 bytes only */
 	key.src_dst[0] = psd[0];
 	key.id = ip_hdr->packet_id;
-- 
2.1.2