[dpdk-dev] [PATCH 0/2] ip_frag: two fixes in reassembly code

[dpdk-dev] [PATCH 0/2] ip_frag: two fixes in reassembly code

[Konstantin Ananyev]

Fix 2 issues reported in reassembly code path.

Konstantin Ananyev (2):
  ip_frag: check fragment length of incoming packet
  ip_frag: use key length for key comparision

 lib/librte_ip_frag/ip_frag_common.h      | 14 ++++--------
 lib/librte_ip_frag/rte_ip_frag.h         | 14 +++++++++---
 lib/librte_ip_frag/rte_ipv4_reassembly.c | 22 +++++++++++++------
 lib/librte_ip_frag/rte_ipv6_reassembly.c | 28 +++++++++++++++++-------
 4 files changed, 50 insertions(+), 28 deletions(-)

-- 
2.17.1

[dpdk-dev] [PATCH 1/2] ip_frag: check fragment length of incoming packet

[Konstantin Ananyev]

Under some conditions ill-formed fragments might cause
reassembly code to corrupt mbufs and/or crash.
Let say the following fragments sequence:
<ofs=0,len=100, flags=MF>
<ofs=96,len=100, flags=MF>
<ofs=200,len=0,flags=MF>
<ofs=200,len=100,flags=0>
can trigger the problem.
To overcome such situation, added check that fragment length
of incoming value is greater than zero.

Reported-by: <ryan.e.hall@intel.com>
Reported-by: <alexander.v.gutkin@intel.com>

Fixes: 601e279df074 ("ip_frag: move fragmentation/reassembly headers into a library")
Fixes: 4f1a8f633862 ("ip_frag: add IPv6 reassembly")
Cc: stable@dpdk.org

Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
---
 lib/librte_ip_frag/rte_ipv4_reassembly.c | 22 +++++++++++++------
 lib/librte_ip_frag/rte_ipv6_reassembly.c | 28 +++++++++++++++++-------
 2 files changed, 35 insertions(+), 15 deletions(-)

diff --git a/lib/librte_ip_frag/rte_ipv4_reassembly.c b/lib/librte_ip_frag/rte_ipv4_reassembly.c
index 4956b99ea..1029b7abc 100644
--- a/lib/librte_ip_frag/rte_ipv4_reassembly.c
+++ b/lib/librte_ip_frag/rte_ipv4_reassembly.c
@@ -36,8 +36,11 @@ ipv4_frag_reassemble(struct ip_frag_pkt *fp)
 			/* previous fragment found. */
 			if(fp->frags[i].ofs + fp->frags[i].len == ofs) {
 
+				RTE_ASSERT(curr_idx != i);
+
 				/* adjust start of the last fragment data. */
-				rte_pktmbuf_adj(m, (uint16_t)(m->l2_len + m->l3_len));
+				rte_pktmbuf_adj(m,
+					(uint16_t)(m->l2_len + m->l3_len));
 				rte_pktmbuf_chain(fp->frags[i].mb, m);
 
 				/* this mbuf should not be accessed directly */
@@ -96,14 +99,14 @@ ipv4_frag_reassemble(struct ip_frag_pkt *fp)
  */
 struct rte_mbuf *
 rte_ipv4_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
-		struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb, uint64_t tms,
-		struct ipv4_hdr *ip_hdr)
+	struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb, uint64_t tms,
+	struct ipv4_hdr *ip_hdr)
 {
 	struct ip_frag_pkt *fp;
 	struct ip_frag_key key;
 	const unaligned_uint64_t *psd;
-	uint16_t ip_len;
 	uint16_t flag_offset, ip_ofs, ip_flag;
+	int32_t ip_len;
 
 	flag_offset = rte_be_to_cpu_16(ip_hdr->fragment_offset);
 	ip_ofs = (uint16_t)(flag_offset & IPV4_HDR_OFFSET_MASK);
@@ -116,12 +119,11 @@ rte_ipv4_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
 	key.key_len = IPV4_KEYLEN;
 
 	ip_ofs *= IPV4_HDR_OFFSET_UNITS;
-	ip_len = (uint16_t)(rte_be_to_cpu_16(ip_hdr->total_length) -
-		mb->l3_len);
+	ip_len = rte_be_to_cpu_16(ip_hdr->total_length) - mb->l3_len;
 
 	IP_FRAG_LOG(DEBUG, "%s:%d:\n"
 		"mbuf: %p, tms: %" PRIu64
-		", key: <%" PRIx64 ", %#x>, ofs: %u, len: %u, flags: %#x\n"
+		", key: <%" PRIx64 ", %#x>, ofs: %u, len: %d, flags: %#x\n"
 		"tbl: %p, max_cycles: %" PRIu64 ", entry_mask: %#x, "
 		"max_entries: %u, use_entries: %u\n\n",
 		__func__, __LINE__,
@@ -129,6 +131,12 @@ rte_ipv4_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
 		tbl, tbl->max_cycles, tbl->entry_mask, tbl->max_entries,
 		tbl->use_entries);
 
+	/* check that fragment length is greater then zero. */
+	if (ip_len <= 0) {
+		IP_FRAG_MBUF2DR(dr, mb);
+		return NULL;
+	}
+
 	/* try to find/add entry into the fragment's table. */
 	if ((fp = ip_frag_find(tbl, dr, &key, tms)) == NULL) {
 		IP_FRAG_MBUF2DR(dr, mb);
diff --git a/lib/librte_ip_frag/rte_ipv6_reassembly.c b/lib/librte_ip_frag/rte_ipv6_reassembly.c
index db249fe60..855e3f740 100644
--- a/lib/librte_ip_frag/rte_ipv6_reassembly.c
+++ b/lib/librte_ip_frag/rte_ipv6_reassembly.c
@@ -59,8 +59,11 @@ ipv6_frag_reassemble(struct ip_frag_pkt *fp)
 			/* previous fragment found. */
 			if (fp->frags[i].ofs + fp->frags[i].len == ofs) {
 
+				RTE_ASSERT(curr_idx != i);
+
 				/* adjust start of the last fragment data. */
-				rte_pktmbuf_adj(m, (uint16_t)(m->l2_len + m->l3_len));
+				rte_pktmbuf_adj(m,
+					(uint16_t)(m->l2_len + m->l3_len));
 				rte_pktmbuf_chain(fp->frags[i].mb, m);
 
 				/* this mbuf should not be accessed directly */
@@ -135,12 +138,13 @@ ipv6_frag_reassemble(struct ip_frag_pkt *fp)
 #define FRAG_OFFSET(x) (rte_cpu_to_be_16(x) >> 3)
 struct rte_mbuf *
 rte_ipv6_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
-		struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb, uint64_t tms,
-		struct ipv6_hdr *ip_hdr, struct ipv6_extension_fragment *frag_hdr)
+	struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb, uint64_t tms,
+	struct ipv6_hdr *ip_hdr, struct ipv6_extension_fragment *frag_hdr)
 {
 	struct ip_frag_pkt *fp;
 	struct ip_frag_key key;
-	uint16_t ip_len, ip_ofs;
+	uint16_t ip_ofs;
+	int32_t ip_len;
 
 	rte_memcpy(&key.src_dst[0], ip_hdr->src_addr, 16);
 	rte_memcpy(&key.src_dst[2], ip_hdr->dst_addr, 16);
@@ -151,15 +155,17 @@ rte_ipv6_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
 	ip_ofs = FRAG_OFFSET(frag_hdr->frag_data) * 8;
 
 	/*
-	 * as per RFC2460, payload length contains all extension headers as well.
-	 * since we don't support anything but frag headers, this is what we remove
-	 * from the payload len.
+	 * as per RFC2460, payload length contains all extension headers
+	 * as well.
+	 * since we don't support anything but frag headers,
+	 * this is what we remove from the payload len.
 	 */
 	ip_len = rte_be_to_cpu_16(ip_hdr->payload_len) - sizeof(*frag_hdr);
 
 	IP_FRAG_LOG(DEBUG, "%s:%d:\n"
 		"mbuf: %p, tms: %" PRIu64
-		", key: <" IPv6_KEY_BYTES_FMT ", %#x>, ofs: %u, len: %u, flags: %#x\n"
+		", key: <" IPv6_KEY_BYTES_FMT ", %#x>, "
+		"ofs: %u, len: %d, flags: %#x\n"
 		"tbl: %p, max_cycles: %" PRIu64 ", entry_mask: %#x, "
 		"max_entries: %u, use_entries: %u\n\n",
 		__func__, __LINE__,
@@ -168,6 +174,12 @@ rte_ipv6_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
 		tbl, tbl->max_cycles, tbl->entry_mask, tbl->max_entries,
 		tbl->use_entries);
 
+	/* check that fragment length is greater then zero. */
+	if (ip_len <= 0) {
+		IP_FRAG_MBUF2DR(dr, mb);
+		return NULL;
+	}
+
 	/* try to find/add entry into the fragment's table. */
 	fp = ip_frag_find(tbl, dr, &key, tms);
 	if (fp == NULL) {
-- 
2.17.1

Re: [dpdk-dev] [dpdk-stable] [PATCH 1/2] ip_frag: check fragment length of incoming packet

[Thomas Monjalon]

05/11/2018 11:26, Konstantin Ananyev:
> Under some conditions ill-formed fragments might cause
> reassembly code to corrupt mbufs and/or crash.
> Let say the following fragments sequence:
> <ofs=0,len=100, flags=MF>
> <ofs=96,len=100, flags=MF>
> <ofs=200,len=0,flags=MF>
> <ofs=200,len=100,flags=0>
> can trigger the problem.
> To overcome such situation, added check that fragment length
> of incoming value is greater than zero.
> 
> Reported-by: <ryan.e.hall@intel.com>
> Reported-by: <alexander.v.gutkin@intel.com>

Please, could you provide the full names?
Note: it is usually inserted just before your Signed-off.

> Fixes: 601e279df074 ("ip_frag: move fragmentation/reassembly headers into a library")
> Fixes: 4f1a8f633862 ("ip_frag: add IPv6 reassembly")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>

[dpdk-dev] [PATCH 2/2] ip_frag: use key length for key comparision

[Konstantin Ananyev]

Right now reassembly code relies on src_dst[] being all zeroes to
determine is it  free/occupied entry in the fragments table.
This is suboptimal and error prone - user can crash DPDK ip_reassembly
app by something like the following scapy script:
x=Ether(src=...,dst=...)/IP(dst='0.0.0.0',src='0.0.0.0',id=0)/('X'*1000)
frags=fragment(x, fragsize=500)
sendp(frags, iface=...)
To overcome that issue and reduce overhead of
'key invalidate'  and 'key is empty' operations -
add key_len into keys comparision procedure.

Reported-by: <ryan.e.hall@intel.com>
Reported-by: <alexander.v.gutkin@intel.com>

Fixes: 4f1a8f633862 ("ip_frag: add IPv6 reassembly")
Cc: stable@dpdk.org

Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
---
 lib/librte_ip_frag/ip_frag_common.h | 14 ++++----------
 lib/librte_ip_frag/rte_ip_frag.h    | 14 +++++++++++---
 2 files changed, 15 insertions(+), 13 deletions(-)

diff --git a/lib/librte_ip_frag/ip_frag_common.h b/lib/librte_ip_frag/ip_frag_common.h
index 0f62e2e16..a17a74076 100644
--- a/lib/librte_ip_frag/ip_frag_common.h
+++ b/lib/librte_ip_frag/ip_frag_common.h
@@ -58,20 +58,14 @@ struct rte_mbuf *ipv6_frag_reassemble(struct ip_frag_pkt *fp);
 static inline int
 ip_frag_key_is_empty(const struct ip_frag_key * key)
 {
-	uint32_t i;
-	for (i = 0; i < RTE_MIN(key->key_len, RTE_DIM(key->src_dst)); i++)
-		if (key->src_dst[i] != 0)
-			return 0;
-	return 1;
+	return (key->key_len == 0);
 }
 
-/* empty the key */
+/* invalidate the key */
 static inline void
 ip_frag_key_invalidate(struct ip_frag_key * key)
 {
-	uint32_t i;
-	for (i = 0; i < key->key_len; i++)
-		key->src_dst[i] = 0;
+	key->key_len = 0;
 }
 
 /* compare two keys */
@@ -80,7 +74,7 @@ ip_frag_key_cmp(const struct ip_frag_key * k1, const struct ip_frag_key * k2)
 {
 	uint32_t i;
 	uint64_t val;
-	val = k1->id ^ k2->id;
+	val = k1->id_key_len ^ k2->id_key_len;
 	for (i = 0; i < k1->key_len; i++)
 		val |= k1->src_dst[i] ^ k2->src_dst[i];
 	return val;
diff --git a/lib/librte_ip_frag/rte_ip_frag.h b/lib/librte_ip_frag/rte_ip_frag.h
index 7f425f610..a4ccaf9d1 100644
--- a/lib/librte_ip_frag/rte_ip_frag.h
+++ b/lib/librte_ip_frag/rte_ip_frag.h
@@ -44,9 +44,17 @@ struct ip_frag {
 
 /** @internal <src addr, dst_addr, id> to uniquely identify fragmented datagram. */
 struct ip_frag_key {
-	uint64_t src_dst[4];      /**< src address, first 8 bytes used for IPv4 */
-	uint32_t id;           /**< dst address */
-	uint32_t key_len;      /**< src/dst key length */
+	uint64_t src_dst[4];
+	/**< src and dst address, only first 8 bytes used for IPv4 */
+	RTE_STD_C11
+	union {
+		uint64_t id_key_len; /**< combined for easy fetch */
+		__extension__
+		struct {
+			uint32_t id;       /**< packet id */
+			uint32_t key_len;  /**< src/dst key length */
+		};
+	};
 };
 
 /**
-- 
2.17.1

[dpdk-dev] [PATCH v2 0/2] ip_frag: two fixes in reassembly code

[Konstantin Ananyev]

Fix 2 issues reported in reassembly code path.

Konstantin Ananyev (2):
  ip_frag: check fragment length of incoming packet
  ip_frag: use key length for key comparision

 lib/librte_ip_frag/ip_frag_common.h      | 14 ++++--------
 lib/librte_ip_frag/rte_ip_frag.h         | 14 +++++++++---
 lib/librte_ip_frag/rte_ipv4_reassembly.c | 22 +++++++++++++------
 lib/librte_ip_frag/rte_ipv6_reassembly.c | 28 +++++++++++++++++-------
 4 files changed, 50 insertions(+), 28 deletions(-)

-- 
2.17.1

[dpdk-dev] [PATCH v2 1/2] ip_frag: check fragment length of incoming packet

[Konstantin Ananyev]

Under some conditions ill-formed fragments might cause
reassembly code to corrupt mbufs and/or crash.
Let say the following fragments sequence:
<ofs=0,len=100, flags=MF>
<ofs=96,len=100, flags=MF>
<ofs=200,len=0,flags=MF>
<ofs=200,len=100,flags=0>
can trigger the problem.
To overcome such situation, added check that fragment length
of incoming value is greater than zero.

Fixes: 601e279df074 ("ip_frag: move fragmentation/reassembly headers into a library")
Fixes: 4f1a8f633862 ("ip_frag: add IPv6 reassembly")
Cc: stable@dpdk.org

Reported-by: Ryan E Hall <ryan.e.hall@intel.com>
Reported-by: Alexander V Gutkin <alexander.v.gutkin@intel.com>
Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
---
 lib/librte_ip_frag/rte_ipv4_reassembly.c | 22 +++++++++++++------
 lib/librte_ip_frag/rte_ipv6_reassembly.c | 28 +++++++++++++++++-------
 2 files changed, 35 insertions(+), 15 deletions(-)

diff --git a/lib/librte_ip_frag/rte_ipv4_reassembly.c b/lib/librte_ip_frag/rte_ipv4_reassembly.c
index 4956b99ea..1029b7abc 100644
--- a/lib/librte_ip_frag/rte_ipv4_reassembly.c
+++ b/lib/librte_ip_frag/rte_ipv4_reassembly.c
@@ -36,8 +36,11 @@ ipv4_frag_reassemble(struct ip_frag_pkt *fp)
 			/* previous fragment found. */
 			if(fp->frags[i].ofs + fp->frags[i].len == ofs) {
 
+				RTE_ASSERT(curr_idx != i);
+
 				/* adjust start of the last fragment data. */
-				rte_pktmbuf_adj(m, (uint16_t)(m->l2_len + m->l3_len));
+				rte_pktmbuf_adj(m,
+					(uint16_t)(m->l2_len + m->l3_len));
 				rte_pktmbuf_chain(fp->frags[i].mb, m);
 
 				/* this mbuf should not be accessed directly */
@@ -96,14 +99,14 @@ ipv4_frag_reassemble(struct ip_frag_pkt *fp)
  */
 struct rte_mbuf *
 rte_ipv4_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
-		struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb, uint64_t tms,
-		struct ipv4_hdr *ip_hdr)
+	struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb, uint64_t tms,
+	struct ipv4_hdr *ip_hdr)
 {
 	struct ip_frag_pkt *fp;
 	struct ip_frag_key key;
 	const unaligned_uint64_t *psd;
-	uint16_t ip_len;
 	uint16_t flag_offset, ip_ofs, ip_flag;
+	int32_t ip_len;
 
 	flag_offset = rte_be_to_cpu_16(ip_hdr->fragment_offset);
 	ip_ofs = (uint16_t)(flag_offset & IPV4_HDR_OFFSET_MASK);
@@ -116,12 +119,11 @@ rte_ipv4_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
 	key.key_len = IPV4_KEYLEN;
 
 	ip_ofs *= IPV4_HDR_OFFSET_UNITS;
-	ip_len = (uint16_t)(rte_be_to_cpu_16(ip_hdr->total_length) -
-		mb->l3_len);
+	ip_len = rte_be_to_cpu_16(ip_hdr->total_length) - mb->l3_len;
 
 	IP_FRAG_LOG(DEBUG, "%s:%d:\n"
 		"mbuf: %p, tms: %" PRIu64
-		", key: <%" PRIx64 ", %#x>, ofs: %u, len: %u, flags: %#x\n"
+		", key: <%" PRIx64 ", %#x>, ofs: %u, len: %d, flags: %#x\n"
 		"tbl: %p, max_cycles: %" PRIu64 ", entry_mask: %#x, "
 		"max_entries: %u, use_entries: %u\n\n",
 		__func__, __LINE__,
@@ -129,6 +131,12 @@ rte_ipv4_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
 		tbl, tbl->max_cycles, tbl->entry_mask, tbl->max_entries,
 		tbl->use_entries);
 
+	/* check that fragment length is greater then zero. */
+	if (ip_len <= 0) {
+		IP_FRAG_MBUF2DR(dr, mb);
+		return NULL;
+	}
+
 	/* try to find/add entry into the fragment's table. */
 	if ((fp = ip_frag_find(tbl, dr, &key, tms)) == NULL) {
 		IP_FRAG_MBUF2DR(dr, mb);
diff --git a/lib/librte_ip_frag/rte_ipv6_reassembly.c b/lib/librte_ip_frag/rte_ipv6_reassembly.c
index db249fe60..855e3f740 100644
--- a/lib/librte_ip_frag/rte_ipv6_reassembly.c
+++ b/lib/librte_ip_frag/rte_ipv6_reassembly.c
@@ -59,8 +59,11 @@ ipv6_frag_reassemble(struct ip_frag_pkt *fp)
 			/* previous fragment found. */
 			if (fp->frags[i].ofs + fp->frags[i].len == ofs) {
 
+				RTE_ASSERT(curr_idx != i);
+
 				/* adjust start of the last fragment data. */
-				rte_pktmbuf_adj(m, (uint16_t)(m->l2_len + m->l3_len));
+				rte_pktmbuf_adj(m,
+					(uint16_t)(m->l2_len + m->l3_len));
 				rte_pktmbuf_chain(fp->frags[i].mb, m);
 
 				/* this mbuf should not be accessed directly */
@@ -135,12 +138,13 @@ ipv6_frag_reassemble(struct ip_frag_pkt *fp)
 #define FRAG_OFFSET(x) (rte_cpu_to_be_16(x) >> 3)
 struct rte_mbuf *
 rte_ipv6_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
-		struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb, uint64_t tms,
-		struct ipv6_hdr *ip_hdr, struct ipv6_extension_fragment *frag_hdr)
+	struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb, uint64_t tms,
+	struct ipv6_hdr *ip_hdr, struct ipv6_extension_fragment *frag_hdr)
 {
 	struct ip_frag_pkt *fp;
 	struct ip_frag_key key;
-	uint16_t ip_len, ip_ofs;
+	uint16_t ip_ofs;
+	int32_t ip_len;
 
 	rte_memcpy(&key.src_dst[0], ip_hdr->src_addr, 16);
 	rte_memcpy(&key.src_dst[2], ip_hdr->dst_addr, 16);
@@ -151,15 +155,17 @@ rte_ipv6_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
 	ip_ofs = FRAG_OFFSET(frag_hdr->frag_data) * 8;
 
 	/*
-	 * as per RFC2460, payload length contains all extension headers as well.
-	 * since we don't support anything but frag headers, this is what we remove
-	 * from the payload len.
+	 * as per RFC2460, payload length contains all extension headers
+	 * as well.
+	 * since we don't support anything but frag headers,
+	 * this is what we remove from the payload len.
 	 */
 	ip_len = rte_be_to_cpu_16(ip_hdr->payload_len) - sizeof(*frag_hdr);
 
 	IP_FRAG_LOG(DEBUG, "%s:%d:\n"
 		"mbuf: %p, tms: %" PRIu64
-		", key: <" IPv6_KEY_BYTES_FMT ", %#x>, ofs: %u, len: %u, flags: %#x\n"
+		", key: <" IPv6_KEY_BYTES_FMT ", %#x>, "
+		"ofs: %u, len: %d, flags: %#x\n"
 		"tbl: %p, max_cycles: %" PRIu64 ", entry_mask: %#x, "
 		"max_entries: %u, use_entries: %u\n\n",
 		__func__, __LINE__,
@@ -168,6 +174,12 @@ rte_ipv6_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
 		tbl, tbl->max_cycles, tbl->entry_mask, tbl->max_entries,
 		tbl->use_entries);
 
+	/* check that fragment length is greater then zero. */
+	if (ip_len <= 0) {
+		IP_FRAG_MBUF2DR(dr, mb);
+		return NULL;
+	}
+
 	/* try to find/add entry into the fragment's table. */
 	fp = ip_frag_find(tbl, dr, &key, tms);
 	if (fp == NULL) {
-- 
2.17.1

[dpdk-dev] [PATCH 2/2] ip_frag: use key length for key comparision

[Konstantin Ananyev]

Right now reassembly code relies on src_dst[] being all zeroes to
determine is it  free/occupied entry in the fragments table.
This is suboptimal and error prone - user can crash DPDK ip_reassembly
app by something like the following scapy script:
x=Ether(src=...,dst=...)/IP(dst='0.0.0.0',src='0.0.0.0',id=0)/('X'*1000)
frags=fragment(x, fragsize=500)
sendp(frags, iface=...)
To overcome that issue and reduce overhead of
'key invalidate'  and 'key is empty' operations -
add key_len into keys comparision procedure.

Fixes: 4f1a8f633862 ("ip_frag: add IPv6 reassembly")
Cc: stable@dpdk.org

Reported-by: Ryan E Hall <ryan.e.hall@intel.com>
Reported-by: Alexander V Gutkin <alexander.v.gutkin@intel.com>
Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
---
 lib/librte_ip_frag/ip_frag_common.h | 14 ++++----------
 lib/librte_ip_frag/rte_ip_frag.h    | 14 +++++++++++---
 2 files changed, 15 insertions(+), 13 deletions(-)

diff --git a/lib/librte_ip_frag/ip_frag_common.h b/lib/librte_ip_frag/ip_frag_common.h
index 0f62e2e16..a17a74076 100644
--- a/lib/librte_ip_frag/ip_frag_common.h
+++ b/lib/librte_ip_frag/ip_frag_common.h
@@ -58,20 +58,14 @@ struct rte_mbuf *ipv6_frag_reassemble(struct ip_frag_pkt *fp);
 static inline int
 ip_frag_key_is_empty(const struct ip_frag_key * key)
 {
-	uint32_t i;
-	for (i = 0; i < RTE_MIN(key->key_len, RTE_DIM(key->src_dst)); i++)
-		if (key->src_dst[i] != 0)
-			return 0;
-	return 1;
+	return (key->key_len == 0);
 }
 
-/* empty the key */
+/* invalidate the key */
 static inline void
 ip_frag_key_invalidate(struct ip_frag_key * key)
 {
-	uint32_t i;
-	for (i = 0; i < key->key_len; i++)
-		key->src_dst[i] = 0;
+	key->key_len = 0;
 }
 
 /* compare two keys */
@@ -80,7 +74,7 @@ ip_frag_key_cmp(const struct ip_frag_key * k1, const struct ip_frag_key * k2)
 {
 	uint32_t i;
 	uint64_t val;
-	val = k1->id ^ k2->id;
+	val = k1->id_key_len ^ k2->id_key_len;
 	for (i = 0; i < k1->key_len; i++)
 		val |= k1->src_dst[i] ^ k2->src_dst[i];
 	return val;
diff --git a/lib/librte_ip_frag/rte_ip_frag.h b/lib/librte_ip_frag/rte_ip_frag.h
index 7f425f610..a4ccaf9d1 100644
--- a/lib/librte_ip_frag/rte_ip_frag.h
+++ b/lib/librte_ip_frag/rte_ip_frag.h
@@ -44,9 +44,17 @@ struct ip_frag {
 
 /** @internal <src addr, dst_addr, id> to uniquely identify fragmented datagram. */
 struct ip_frag_key {
-	uint64_t src_dst[4];      /**< src address, first 8 bytes used for IPv4 */
-	uint32_t id;           /**< dst address */
-	uint32_t key_len;      /**< src/dst key length */
+	uint64_t src_dst[4];
+	/**< src and dst address, only first 8 bytes used for IPv4 */
+	RTE_STD_C11
+	union {
+		uint64_t id_key_len; /**< combined for easy fetch */
+		__extension__
+		struct {
+			uint32_t id;       /**< packet id */
+			uint32_t key_len;  /**< src/dst key length */
+		};
+	};
 };
 
 /**
-- 
2.17.1

[dpdk-dev] [PATCH v2 0/2] ip_frag: two fixes in reassembly code

[Konstantin Ananyev]

Fix 2 issues reported in reassembly code path.

Konstantin Ananyev (2):
  ip_frag: check fragment length of incoming packet
  ip_frag: use key length for key comparision

 lib/librte_ip_frag/ip_frag_common.h      | 14 ++++--------
 lib/librte_ip_frag/rte_ip_frag.h         | 14 +++++++++---
 lib/librte_ip_frag/rte_ipv4_reassembly.c | 22 +++++++++++++------
 lib/librte_ip_frag/rte_ipv6_reassembly.c | 28 +++++++++++++++++-------
 4 files changed, 50 insertions(+), 28 deletions(-)

-- 
2.17.1

Re: [dpdk-dev] [PATCH v2 0/2] ip_frag: two fixes in reassembly code

[Thomas Monjalon]

05/11/2018 13:18, Konstantin Ananyev:
> Fix 2 issues reported in reassembly code path.
> 
> Konstantin Ananyev (2):
>   ip_frag: check fragment length of incoming packet
>   ip_frag: use key length for key comparision

Applied, thanks

[dpdk-dev] [PATCH v2 1/2] ip_frag: check fragment length of incoming packet

[Konstantin Ananyev]

Under some conditions ill-formed fragments might cause
reassembly code to corrupt mbufs and/or crash.
Let say the following fragments sequence:
<ofs=0,len=100, flags=MF>
<ofs=96,len=100, flags=MF>
<ofs=200,len=0,flags=MF>
<ofs=200,len=100,flags=0>
can trigger the problem.
To overcome such situation, added check that fragment length
of incoming value is greater than zero.

Fixes: 601e279df074 ("ip_frag: move fragmentation/reassembly headers into a library")
Fixes: 4f1a8f633862 ("ip_frag: add IPv6 reassembly")
Cc: stable@dpdk.org

Reported-by: Ryan E Hall <ryan.e.hall@intel.com>
Reported-by: Alexander V Gutkin <alexander.v.gutkin@intel.com>
Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
---
 lib/librte_ip_frag/rte_ipv4_reassembly.c | 22 +++++++++++++------
 lib/librte_ip_frag/rte_ipv6_reassembly.c | 28 +++++++++++++++++-------
 2 files changed, 35 insertions(+), 15 deletions(-)

diff --git a/lib/librte_ip_frag/rte_ipv4_reassembly.c b/lib/librte_ip_frag/rte_ipv4_reassembly.c
index 4956b99ea..1029b7abc 100644
--- a/lib/librte_ip_frag/rte_ipv4_reassembly.c
+++ b/lib/librte_ip_frag/rte_ipv4_reassembly.c
@@ -36,8 +36,11 @@ ipv4_frag_reassemble(struct ip_frag_pkt *fp)
 			/* previous fragment found. */
 			if(fp->frags[i].ofs + fp->frags[i].len == ofs) {
 
+				RTE_ASSERT(curr_idx != i);
+
 				/* adjust start of the last fragment data. */
-				rte_pktmbuf_adj(m, (uint16_t)(m->l2_len + m->l3_len));
+				rte_pktmbuf_adj(m,
+					(uint16_t)(m->l2_len + m->l3_len));
 				rte_pktmbuf_chain(fp->frags[i].mb, m);
 
 				/* this mbuf should not be accessed directly */
@@ -96,14 +99,14 @@ ipv4_frag_reassemble(struct ip_frag_pkt *fp)
  */
 struct rte_mbuf *
 rte_ipv4_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
-		struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb, uint64_t tms,
-		struct ipv4_hdr *ip_hdr)
+	struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb, uint64_t tms,
+	struct ipv4_hdr *ip_hdr)
 {
 	struct ip_frag_pkt *fp;
 	struct ip_frag_key key;
 	const unaligned_uint64_t *psd;
-	uint16_t ip_len;
 	uint16_t flag_offset, ip_ofs, ip_flag;
+	int32_t ip_len;
 
 	flag_offset = rte_be_to_cpu_16(ip_hdr->fragment_offset);
 	ip_ofs = (uint16_t)(flag_offset & IPV4_HDR_OFFSET_MASK);
@@ -116,12 +119,11 @@ rte_ipv4_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
 	key.key_len = IPV4_KEYLEN;
 
 	ip_ofs *= IPV4_HDR_OFFSET_UNITS;
-	ip_len = (uint16_t)(rte_be_to_cpu_16(ip_hdr->total_length) -
-		mb->l3_len);
+	ip_len = rte_be_to_cpu_16(ip_hdr->total_length) - mb->l3_len;
 
 	IP_FRAG_LOG(DEBUG, "%s:%d:\n"
 		"mbuf: %p, tms: %" PRIu64
-		", key: <%" PRIx64 ", %#x>, ofs: %u, len: %u, flags: %#x\n"
+		", key: <%" PRIx64 ", %#x>, ofs: %u, len: %d, flags: %#x\n"
 		"tbl: %p, max_cycles: %" PRIu64 ", entry_mask: %#x, "
 		"max_entries: %u, use_entries: %u\n\n",
 		__func__, __LINE__,
@@ -129,6 +131,12 @@ rte_ipv4_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
 		tbl, tbl->max_cycles, tbl->entry_mask, tbl->max_entries,
 		tbl->use_entries);
 
+	/* check that fragment length is greater then zero. */
+	if (ip_len <= 0) {
+		IP_FRAG_MBUF2DR(dr, mb);
+		return NULL;
+	}
+
 	/* try to find/add entry into the fragment's table. */
 	if ((fp = ip_frag_find(tbl, dr, &key, tms)) == NULL) {
 		IP_FRAG_MBUF2DR(dr, mb);
diff --git a/lib/librte_ip_frag/rte_ipv6_reassembly.c b/lib/librte_ip_frag/rte_ipv6_reassembly.c
index db249fe60..855e3f740 100644
--- a/lib/librte_ip_frag/rte_ipv6_reassembly.c
+++ b/lib/librte_ip_frag/rte_ipv6_reassembly.c
@@ -59,8 +59,11 @@ ipv6_frag_reassemble(struct ip_frag_pkt *fp)
 			/* previous fragment found. */
 			if (fp->frags[i].ofs + fp->frags[i].len == ofs) {
 
+				RTE_ASSERT(curr_idx != i);
+
 				/* adjust start of the last fragment data. */
-				rte_pktmbuf_adj(m, (uint16_t)(m->l2_len + m->l3_len));
+				rte_pktmbuf_adj(m,
+					(uint16_t)(m->l2_len + m->l3_len));
 				rte_pktmbuf_chain(fp->frags[i].mb, m);
 
 				/* this mbuf should not be accessed directly */
@@ -135,12 +138,13 @@ ipv6_frag_reassemble(struct ip_frag_pkt *fp)
 #define FRAG_OFFSET(x) (rte_cpu_to_be_16(x) >> 3)
 struct rte_mbuf *
 rte_ipv6_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
-		struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb, uint64_t tms,
-		struct ipv6_hdr *ip_hdr, struct ipv6_extension_fragment *frag_hdr)
+	struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb, uint64_t tms,
+	struct ipv6_hdr *ip_hdr, struct ipv6_extension_fragment *frag_hdr)
 {
 	struct ip_frag_pkt *fp;
 	struct ip_frag_key key;
-	uint16_t ip_len, ip_ofs;
+	uint16_t ip_ofs;
+	int32_t ip_len;
 
 	rte_memcpy(&key.src_dst[0], ip_hdr->src_addr, 16);
 	rte_memcpy(&key.src_dst[2], ip_hdr->dst_addr, 16);
@@ -151,15 +155,17 @@ rte_ipv6_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
 	ip_ofs = FRAG_OFFSET(frag_hdr->frag_data) * 8;
 
 	/*
-	 * as per RFC2460, payload length contains all extension headers as well.
-	 * since we don't support anything but frag headers, this is what we remove
-	 * from the payload len.
+	 * as per RFC2460, payload length contains all extension headers
+	 * as well.
+	 * since we don't support anything but frag headers,
+	 * this is what we remove from the payload len.
 	 */
 	ip_len = rte_be_to_cpu_16(ip_hdr->payload_len) - sizeof(*frag_hdr);
 
 	IP_FRAG_LOG(DEBUG, "%s:%d:\n"
 		"mbuf: %p, tms: %" PRIu64
-		", key: <" IPv6_KEY_BYTES_FMT ", %#x>, ofs: %u, len: %u, flags: %#x\n"
+		", key: <" IPv6_KEY_BYTES_FMT ", %#x>, "
+		"ofs: %u, len: %d, flags: %#x\n"
 		"tbl: %p, max_cycles: %" PRIu64 ", entry_mask: %#x, "
 		"max_entries: %u, use_entries: %u\n\n",
 		__func__, __LINE__,
@@ -168,6 +174,12 @@ rte_ipv6_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
 		tbl, tbl->max_cycles, tbl->entry_mask, tbl->max_entries,
 		tbl->use_entries);
 
+	/* check that fragment length is greater then zero. */
+	if (ip_len <= 0) {
+		IP_FRAG_MBUF2DR(dr, mb);
+		return NULL;
+	}
+
 	/* try to find/add entry into the fragment's table. */
 	fp = ip_frag_find(tbl, dr, &key, tms);
 	if (fp == NULL) {
-- 
2.17.1

[dpdk-dev] [PATCH v2 2/2] ip_frag: use key length for key comparision

[Konstantin Ananyev]

Right now reassembly code relies on src_dst[] being all zeroes to
determine is it  free/occupied entry in the fragments table.
This is suboptimal and error prone - user can crash DPDK ip_reassembly
app by something like the following scapy script:
x=Ether(src=...,dst=...)/IP(dst='0.0.0.0',src='0.0.0.0',id=0)/('X'*1000)
frags=fragment(x, fragsize=500)
sendp(frags, iface=...)
To overcome that issue and reduce overhead of
'key invalidate'  and 'key is empty' operations -
add key_len into keys comparision procedure.

Fixes: 4f1a8f633862 ("ip_frag: add IPv6 reassembly")
Cc: stable@dpdk.org

Reported-by: Ryan E Hall <ryan.e.hall@intel.com>
Reported-by: Alexander V Gutkin <alexander.v.gutkin@intel.com>
Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
---
 lib/librte_ip_frag/ip_frag_common.h | 14 ++++----------
 lib/librte_ip_frag/rte_ip_frag.h    | 14 +++++++++++---
 2 files changed, 15 insertions(+), 13 deletions(-)

diff --git a/lib/librte_ip_frag/ip_frag_common.h b/lib/librte_ip_frag/ip_frag_common.h
index 0f62e2e16..a17a74076 100644
--- a/lib/librte_ip_frag/ip_frag_common.h
+++ b/lib/librte_ip_frag/ip_frag_common.h
@@ -58,20 +58,14 @@ struct rte_mbuf *ipv6_frag_reassemble(struct ip_frag_pkt *fp);
 static inline int
 ip_frag_key_is_empty(const struct ip_frag_key * key)
 {
-	uint32_t i;
-	for (i = 0; i < RTE_MIN(key->key_len, RTE_DIM(key->src_dst)); i++)
-		if (key->src_dst[i] != 0)
-			return 0;
-	return 1;
+	return (key->key_len == 0);
 }
 
-/* empty the key */
+/* invalidate the key */
 static inline void
 ip_frag_key_invalidate(struct ip_frag_key * key)
 {
-	uint32_t i;
-	for (i = 0; i < key->key_len; i++)
-		key->src_dst[i] = 0;
+	key->key_len = 0;
 }
 
 /* compare two keys */
@@ -80,7 +74,7 @@ ip_frag_key_cmp(const struct ip_frag_key * k1, const struct ip_frag_key * k2)
 {
 	uint32_t i;
 	uint64_t val;
-	val = k1->id ^ k2->id;
+	val = k1->id_key_len ^ k2->id_key_len;
 	for (i = 0; i < k1->key_len; i++)
 		val |= k1->src_dst[i] ^ k2->src_dst[i];
 	return val;
diff --git a/lib/librte_ip_frag/rte_ip_frag.h b/lib/librte_ip_frag/rte_ip_frag.h
index 7f425f610..a4ccaf9d1 100644
--- a/lib/librte_ip_frag/rte_ip_frag.h
+++ b/lib/librte_ip_frag/rte_ip_frag.h
@@ -44,9 +44,17 @@ struct ip_frag {
 
 /** @internal <src addr, dst_addr, id> to uniquely identify fragmented datagram. */
 struct ip_frag_key {
-	uint64_t src_dst[4];      /**< src address, first 8 bytes used for IPv4 */
-	uint32_t id;           /**< dst address */
-	uint32_t key_len;      /**< src/dst key length */
+	uint64_t src_dst[4];
+	/**< src and dst address, only first 8 bytes used for IPv4 */
+	RTE_STD_C11
+	union {
+		uint64_t id_key_len; /**< combined for easy fetch */
+		__extension__
+		struct {
+			uint32_t id;       /**< packet id */
+			uint32_t key_len;  /**< src/dst key length */
+		};
+	};
 };
 
 /**
-- 
2.17.1

Re: [dpdk-dev] [PATCH v2 2/2] ip_frag: use key length for key comparision

[Burakov, Anatoly]

On 05-Nov-18 12:18 PM, Konstantin Ananyev wrote:
> Right now reassembly code relies on src_dst[] being all zeroes to
> determine is it  free/occupied entry in the fragments table.
> This is suboptimal and error prone - user can crash DPDK ip_reassembly
> app by something like the following scapy script:
> x=Ether(src=...,dst=...)/IP(dst='0.0.0.0',src='0.0.0.0',id=0)/('X'*1000)
> frags=fragment(x, fragsize=500)
> sendp(frags, iface=...)
> To overcome that issue and reduce overhead of
> 'key invalidate'  and 'key is empty' operations -
> add key_len into keys comparision procedure.
> 
> Fixes: 4f1a8f633862 ("ip_frag: add IPv6 reassembly")
> Cc: stable@dpdk.org
> 
> Reported-by: Ryan E Hall <ryan.e.hall@intel.com>
> Reported-by: Alexander V Gutkin <alexander.v.gutkin@intel.com>
> Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
> ---



> @@ -44,9 +44,17 @@ struct ip_frag {
>   
>   /** @internal <src addr, dst_addr, id> to uniquely identify fragmented datagram. */
>   struct ip_frag_key {
> -	uint64_t src_dst[4];      /**< src address, first 8 bytes used for IPv4 */
> -	uint32_t id;           /**< dst address */
> -	uint32_t key_len;      /**< src/dst key length */
> +	uint64_t src_dst[4];
> +	/**< src and dst address, only first 8 bytes used for IPv4 */
> +	RTE_STD_C11
> +	union {
> +		uint64_t id_key_len; /**< combined for easy fetch */
> +		__extension__
> +		struct {
> +			uint32_t id;       /**< packet id */
> +			uint32_t key_len;  /**< src/dst key length */
> +		};
> +	};
>   };

Would that break ABI?

>   
>   /**
> 


-- 
Thanks,
Anatoly

Re: [dpdk-dev] [PATCH v2 2/2] ip_frag: use key length for key comparision

[Ananyev, Konstantin]

> -----Original Message-----
> From: Burakov, Anatoly
> Sent: Tuesday, November 6, 2018 10:54 AM
> To: Ananyev, Konstantin <konstantin.ananyev@intel.com>; dev@dpdk.org
> Cc: stable@dpdk.org; Hall, Ryan E <ryan.e.hall@intel.com>; Gutkin, Alexander V <alexander.v.gutkin@intel.com>
> Subject: Re: [dpdk-dev] [PATCH v2 2/2] ip_frag: use key length for key comparision
> 
> On 05-Nov-18 12:18 PM, Konstantin Ananyev wrote:
> > Right now reassembly code relies on src_dst[] being all zeroes to
> > determine is it  free/occupied entry in the fragments table.
> > This is suboptimal and error prone - user can crash DPDK ip_reassembly
> > app by something like the following scapy script:
> > x=Ether(src=...,dst=...)/IP(dst='0.0.0.0',src='0.0.0.0',id=0)/('X'*1000)
> > frags=fragment(x, fragsize=500)
> > sendp(frags, iface=...)
> > To overcome that issue and reduce overhead of
> > 'key invalidate'  and 'key is empty' operations -
> > add key_len into keys comparision procedure.
> >
> > Fixes: 4f1a8f633862 ("ip_frag: add IPv6 reassembly")
> > Cc: stable@dpdk.org
> >
> > Reported-by: Ryan E Hall <ryan.e.hall@intel.com>
> > Reported-by: Alexander V Gutkin <alexander.v.gutkin@intel.com>
> > Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
> > ---
> 
> 
> 
> > @@ -44,9 +44,17 @@ struct ip_frag {
> >
> >   /** @internal <src addr, dst_addr, id> to uniquely identify fragmented datagram. */
> >   struct ip_frag_key {
> > -	uint64_t src_dst[4];      /**< src address, first 8 bytes used for IPv4 */
> > -	uint32_t id;           /**< dst address */
> > -	uint32_t key_len;      /**< src/dst key length */
> > +	uint64_t src_dst[4];
> > +	/**< src and dst address, only first 8 bytes used for IPv4 */
> > +	RTE_STD_C11
> > +	union {
> > +		uint64_t id_key_len; /**< combined for easy fetch */
> > +		__extension__
> > +		struct {
> > +			uint32_t id;       /**< packet id */
> > +			uint32_t key_len;  /**< src/dst key length */
> > +		};
> > +	};
> >   };
> 
> Would that break ABI?

No, size and layout of the structure  remains the same.
Konstantin