summaryrefslogtreecommitdiffstats
path: root/vnet
diff options
context:
space:
mode:
Diffstat (limited to 'vnet')
-rw-r--r--vnet/Makefile.am1
-rw-r--r--vnet/vnet/ip/icmp4.c258
-rw-r--r--vnet/vnet/ip/icmp4.h48
-rw-r--r--vnet/vnet/ip/icmp6.c5
-rw-r--r--vnet/vnet/ip/ip.h1
-rw-r--r--vnet/vnet/ip/ip4_input.c39
-rw-r--r--vnet/vnet/ip/ip6_input.c10
-rw-r--r--vnet/vnet/ip/ip_frag.c33
-rw-r--r--vnet/vnet/ip/ip_frag.h7
-rw-r--r--vnet/vnet/map/ip4_map.c54
-rw-r--r--vnet/vnet/map/map.c72
-rw-r--r--vnet/vnet/map/map.h31
12 files changed, 368 insertions, 191 deletions
diff --git a/vnet/Makefile.am b/vnet/Makefile.am
index 1bde1e8082c..a0c20e9eb37 100644
--- a/vnet/Makefile.am
+++ b/vnet/Makefile.am
@@ -252,6 +252,7 @@ libvnet_la_SOURCES += \
nobase_include_HEADERS += \
vnet/ip/format.h \
vnet/ip/icmp46_packet.h \
+ vnet/ip/icmp4.h \
vnet/ip/icmp6.h \
vnet/ip/igmp_packet.h \
vnet/ip/ip.h \
diff --git a/vnet/vnet/ip/icmp4.c b/vnet/vnet/ip/icmp4.c
index 4ee8f15df1f..abad5bd9efc 100644
--- a/vnet/vnet/ip/icmp4.c
+++ b/vnet/vnet/ip/icmp4.c
@@ -41,6 +41,13 @@
#include <vnet/ip/ip.h>
#include <vnet/pg/pg.h>
+
+static char * icmp_error_strings[] = {
+#define _(f,s) s,
+ foreach_icmp4_error
+#undef _
+};
+
static u8 * format_ip4_icmp_type_and_code (u8 * s, va_list * args)
{
icmp4_type_t type = va_arg (*args, int);
@@ -96,10 +103,6 @@ static u8 * format_ip4_icmp_header (u8 * s, va_list * args)
return s;
}
-typedef struct {
- u8 packet_data[64];
-} icmp_input_trace_t;
-
static u8 * format_icmp_input_trace (u8 * s, va_list * va)
{
CLIB_UNUSED (vlib_main_t * vm) = va_arg (*va, vlib_main_t *);
@@ -114,20 +117,6 @@ static u8 * format_icmp_input_trace (u8 * s, va_list * va)
}
typedef enum {
- ICMP4_ERROR_UNKNOWN_TYPE,
- ICMP4_ERROR_ECHO_REPLIES_SENT,
- ICMP4_ERROR_TTL_EXPIRE_RESP_SENT,
- ICMP4_ERROR_TTL_EXPIRE_RESP_DROP,
-} icmp_error_t;
-
-static char * icmp_error_strings[] = {
- [ICMP4_ERROR_UNKNOWN_TYPE] = "unknown type",
- [ICMP4_ERROR_ECHO_REPLIES_SENT] = "echo replies sent",
- [ICMP4_ERROR_TTL_EXPIRE_RESP_SENT] = "TTL time exceeded response sent",
- [ICMP4_ERROR_TTL_EXPIRE_RESP_DROP] = "TTL time exceeded response dropped",
-};
-
-typedef enum {
ICMP_INPUT_NEXT_ERROR,
ICMP_INPUT_N_NEXT,
} icmp_input_next_t;
@@ -418,19 +407,42 @@ VLIB_REGISTER_NODE (ip4_icmp_echo_request_node,static) = {
};
typedef enum {
- ICMP4_TTL_EXPIRE_NEXT_DROP,
- ICMP4_TTL_EXPIRE_NEXT_LOOKUP,
- ICMP4_TTL_EXPIRE_N_NEXT,
-} icmp_ttl_expire_next_t;
+ IP4_ICMP_ERROR_NEXT_DROP,
+ IP4_ICMP_ERROR_NEXT_LOOKUP,
+ IP4_ICMP_ERROR_N_NEXT,
+} ip4_icmp_error_next_t;
+
+void
+icmp4_error_set_vnet_buffer (vlib_buffer_t *b, u8 type, u8 code, u32 data)
+{
+ vnet_buffer(b)->ip.icmp.type = type;
+ vnet_buffer(b)->ip.icmp.code = code;
+ vnet_buffer(b)->ip.icmp.data = data;
+}
+
+static u8
+icmp4_icmp_type_to_error (u8 type)
+{
+ switch (type) {
+ case ICMP4_destination_unreachable:
+ return ICMP4_ERROR_DEST_UNREACH_SENT;
+ case ICMP4_time_exceeded:
+ return ICMP4_ERROR_TTL_EXPIRE_SENT;
+ case ICMP4_parameter_problem:
+ return ICMP4_ERROR_PARAM_PROBLEM_SENT;
+ default:
+ return ICMP4_ERROR_DROP;
+ }
+}
static uword
-ip4_icmp_ttl_expire (vlib_main_t * vm,
- vlib_node_runtime_t * node,
- vlib_frame_t * frame)
+ip4_icmp_error (vlib_main_t * vm,
+ vlib_node_runtime_t * node,
+ vlib_frame_t * frame)
{
u32 * from, * to_next;
uword n_left_from, n_left_to_next;
- icmp_ttl_expire_next_t next_index;
+ ip4_icmp_error_next_t next_index;
ip4_main_t *im = &ip4_main;
ip_lookup_main_t * lm = &im->lookup_main;
@@ -442,117 +454,113 @@ ip4_icmp_ttl_expire (vlib_main_t * vm,
vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors,
/* stride */ 1, sizeof (icmp_input_trace_t));
- while (n_left_from > 0)
- {
- vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next);
-
- while (n_left_from > 0 && n_left_to_next > 0)
- {
- u32 pi0 = from[0];
- u32 next0 = ICMP4_TTL_EXPIRE_NEXT_LOOKUP;
- u8 error0 = ICMP4_ERROR_TTL_EXPIRE_RESP_SENT;
- u32 len0, new_len0;
- vlib_buffer_t * p0;
- ip4_header_t * ip0, * out_ip0;
- icmp46_header_t * icmp0;
- ip_csum_t sum;
- u32 sw_if_index0, if_add_index0;
-
- /* Speculatively enqueue p0 to the current next frame */
- to_next[0] = pi0;
- from += 1;
- to_next += 1;
- n_left_from -= 1;
- n_left_to_next -= 1;
-
- p0 = vlib_get_buffer(vm, pi0);
- ip0 = vlib_buffer_get_current(p0);
- len0 = vlib_buffer_length_in_chain (vm, p0);
- sw_if_index0 = vnet_buffer(p0)->sw_if_index[VLIB_RX];
-
- /* Cut payload to just IP header plus first 8 bytes */
- new_len0 = (ip0->ip_version_and_header_length &0xf)*4 + 8;
- if (len0 > new_len0)
- {
- p0->current_length = new_len0; /* should fit in 1st buffer */
- if (PREDICT_FALSE(p0->total_length_not_including_first_buffer))
- { /* clear current_length of all other buffers in chain */
- vlib_buffer_t *b = p0;
- p0->total_length_not_including_first_buffer = 0;
- while (b->flags & VLIB_BUFFER_NEXT_PRESENT)
- {
- b = vlib_get_buffer (vm, b->next_buffer);
- b->current_length = 0;
- }
- }
- }
+ while (n_left_from > 0) {
+ vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next);
- /* Add IP header and ICMP header including a 4 byte unused field */
- vlib_buffer_advance(p0,
- -sizeof(ip4_header_t)-sizeof(icmp46_header_t)-4);
- out_ip0 = vlib_buffer_get_current(p0);
- icmp0 = (icmp46_header_t *) &out_ip0[1];
-
- /* Fill ip header fields */
- out_ip0->ip_version_and_header_length = 0x45;
- out_ip0->tos = 0;
- out_ip0->length = clib_host_to_net_u16(p0->current_length);
- out_ip0->fragment_id = 0;
- out_ip0->ttl = 0xff;
- out_ip0->protocol = IP_PROTOCOL_ICMP;
- out_ip0->dst_address = ip0->src_address;
- if_add_index0 =
- lm->if_address_pool_index_by_sw_if_index[sw_if_index0];
- if (PREDICT_TRUE(if_add_index0 != ~0))
- {
- ip_interface_address_t *if_add =
- pool_elt_at_index(lm->if_address_pool, if_add_index0);
- ip4_address_t *if_ip =
- ip_interface_address_get_address(lm, if_add);
- out_ip0->src_address = *if_ip;
- vlib_error_count (vm, node->node_index, error0, 1);
- }
- else /* interface has no IP4 address - should not happen */
- {
- next0 = ICMP4_TTL_EXPIRE_NEXT_DROP;
- error0 = ICMP4_ERROR_TTL_EXPIRE_RESP_DROP;
- }
- out_ip0->checksum = ip4_header_checksum(out_ip0);
-
- /* Fill icmp header fields */
- icmp0->type = ICMP4_time_exceeded;
- icmp0->code = ICMP4_time_exceeded_ttl_exceeded_in_transit;
- icmp0->checksum = 0;
- sum = ip_incremental_checksum(
- 0, icmp0, p0->current_length - sizeof(ip4_header_t));
- icmp0->checksum = ~ip_csum_fold(sum);
-
- /* Update error status */
- p0->error = node->errors[error0];
-
- /* Verify speculative enqueue, maybe switch current next frame */
- vlib_validate_buffer_enqueue_x1(vm, node, next_index,
- to_next, n_left_to_next,
- pi0, next0);
- }
- vlib_put_next_frame(vm, node, next_index, n_left_to_next);
+ while (n_left_from > 0 && n_left_to_next > 0) {
+ u32 pi0 = from[0];
+ u32 next0 = IP4_ICMP_ERROR_NEXT_LOOKUP;
+ u8 error0 = ICMP4_ERROR_NONE;
+ vlib_buffer_t * p0;
+ ip4_header_t * ip0, * out_ip0;
+ icmp46_header_t * icmp0;
+ u32 sw_if_index0, if_add_index0;
+ ip_csum_t sum;
+
+ /* Speculatively enqueue p0 to the current next frame */
+ to_next[0] = pi0;
+ from += 1;
+ to_next += 1;
+ n_left_from -= 1;
+ n_left_to_next -= 1;
+
+ p0 = vlib_get_buffer(vm, pi0);
+ ip0 = vlib_buffer_get_current(p0);
+ sw_if_index0 = vnet_buffer(p0)->sw_if_index[VLIB_RX];
+
+ /*
+ * RFC1812 says to keep as much of the original packet as
+ * possible within the minimum MTU (576). We cheat "a little"
+ * here by keeping whatever fits in the first buffer, to be more
+ * efficient
+ */
+ if (PREDICT_FALSE(p0->total_length_not_including_first_buffer)) {
+ /* clear current_length of all other buffers in chain */
+ vlib_buffer_t *b = p0;
+ p0->total_length_not_including_first_buffer = 0;
+ while (b->flags & VLIB_BUFFER_NEXT_PRESENT) {
+ b = vlib_get_buffer (vm, b->next_buffer);
+ b->current_length = 0;
+ }
+ }
+ p0->current_length = p0->current_length > 576 ? 576 : p0->current_length;
+
+ /* Add IP header and ICMPv4 header including a 4 byte data field */
+ vlib_buffer_advance(p0,
+ -sizeof(ip4_header_t)-sizeof(icmp46_header_t)-4);
+ out_ip0 = vlib_buffer_get_current(p0);
+ icmp0 = (icmp46_header_t *) &out_ip0[1];
+
+ /* Fill ip header fields */
+ out_ip0->ip_version_and_header_length = 0x45;
+ out_ip0->tos = 0;
+ out_ip0->length = clib_host_to_net_u16(p0->current_length);
+ out_ip0->fragment_id = 0;
+ out_ip0->flags_and_fragment_offset = 0;
+ out_ip0->ttl = 0xff;
+ out_ip0->protocol = IP_PROTOCOL_ICMP;
+ out_ip0->dst_address = ip0->src_address;
+ if_add_index0 =
+ lm->if_address_pool_index_by_sw_if_index[sw_if_index0];
+ if (PREDICT_TRUE(if_add_index0 != ~0)) {
+ ip_interface_address_t *if_add =
+ pool_elt_at_index(lm->if_address_pool, if_add_index0);
+ ip4_address_t *if_ip =
+ ip_interface_address_get_address(lm, if_add);
+ out_ip0->src_address = *if_ip;
+ } else {
+ /* interface has no IP4 address - should not happen */
+ next0 = IP4_ICMP_ERROR_NEXT_DROP;
+ error0 = ICMP4_ERROR_DROP;
+ }
+ out_ip0->checksum = ip4_header_checksum(out_ip0);
+
+ /* Fill icmp header fields */
+ icmp0->type = vnet_buffer(p0)->ip.icmp.type;
+ icmp0->code = vnet_buffer(p0)->ip.icmp.code;
+ *((u32 *)(icmp0 + 1)) = clib_host_to_net_u32(vnet_buffer(p0)->ip.icmp.data);
+ icmp0->checksum = 0;
+ sum = ip_incremental_checksum(0, icmp0, p0->current_length - sizeof(ip4_header_t));
+ icmp0->checksum = ~ip_csum_fold(sum);
+
+ /* Update error status */
+ if (error0 == ICMP4_ERROR_NONE)
+ error0 = icmp4_icmp_type_to_error(icmp0->type);
+ vlib_error_count(vm, node->node_index, error0, 1);
+
+ /* Verify speculative enqueue, maybe switch current next frame */
+ vlib_validate_buffer_enqueue_x1(vm, node, next_index,
+ to_next, n_left_to_next,
+ pi0, next0);
}
+ vlib_put_next_frame(vm, node, next_index, n_left_to_next);
+ }
return frame->n_vectors;
}
-VLIB_REGISTER_NODE (ip4_icmp_ttl_expire_node) = {
- .function = ip4_icmp_ttl_expire,
- .name = "ip4-icmp-ttl-expire",
+VLIB_REGISTER_NODE (ip4_icmp_error_node) = {
+ .function = ip4_icmp_error,
+ .name = "ip4-icmp-error",
.vector_size = sizeof (u32),
.n_errors = ARRAY_LEN (icmp_error_strings),
.error_strings = icmp_error_strings,
- .n_next_nodes = ICMP4_TTL_EXPIRE_N_NEXT,
+ .n_next_nodes = IP4_ICMP_ERROR_N_NEXT,
.next_nodes = {
- [ICMP4_TTL_EXPIRE_NEXT_DROP] = "error-drop",
- [ICMP4_TTL_EXPIRE_NEXT_LOOKUP] = "ip4-lookup",
+ [IP4_ICMP_ERROR_NEXT_DROP] = "error-drop",
+ [IP4_ICMP_ERROR_NEXT_LOOKUP] = "ip4-lookup",
},
.format_trace = format_icmp_input_trace,
diff --git a/vnet/vnet/ip/icmp4.h b/vnet/vnet/ip/icmp4.h
new file mode 100644
index 00000000000..f99bf2dafc3
--- /dev/null
+++ b/vnet/vnet/ip/icmp4.h
@@ -0,0 +1,48 @@
+/*
+ * Copyright (c) 2015 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#ifndef included_vnet_icmp4_h
+#define included_vnet_icmp4_h
+
+#define foreach_icmp4_error \
+ _ (NONE, "valid packets") \
+ _ (UNKNOWN_TYPE, "unknown type") \
+ _ (INVALID_CODE_FOR_TYPE, "invalid code for type") \
+ _ (INVALID_HOP_LIMIT_FOR_TYPE, "hop_limit != 255") \
+ _ (LENGTH_TOO_SMALL_FOR_TYPE, "payload length too small for type") \
+ _ (OPTIONS_WITH_ODD_LENGTH, \
+ "total option length not multiple of 8 bytes") \
+ _ (OPTION_WITH_ZERO_LENGTH, "option has zero length") \
+ _ (ECHO_REPLIES_SENT, "echo replies sent") \
+ _ (DST_LOOKUP_MISS, "icmp6 dst address lookup misses") \
+ _ (DEST_UNREACH_SENT, "destination unreachable response sent") \
+ _ (TTL_EXPIRE_SENT, "hop limit exceeded response sent") \
+ _ (PARAM_PROBLEM_SENT, "parameter Pproblem response sent") \
+ _ (DROP, "error message dropped")
+
+typedef enum {
+#define _(f,s) ICMP4_ERROR_##f,
+ foreach_icmp4_error
+#undef _
+} icmp4_error_t;
+
+typedef struct {
+ u8 packet_data[64];
+} icmp_input_trace_t;
+
+format_function_t format_icmp4_input_trace;
+void ip4_icmp_register_type (vlib_main_t * vm, icmp4_type_t type, u32 node_index);
+void icmp4_error_set_vnet_buffer (vlib_buffer_t *b, u8 type, u8 code, u32 data);
+
+#endif /* included_vnet_icmp4_h */
diff --git a/vnet/vnet/ip/icmp6.c b/vnet/vnet/ip/icmp6.c
index c5eb0f609b8..e6022adfc49 100644
--- a/vnet/vnet/ip/icmp6.c
+++ b/vnet/vnet/ip/icmp6.c
@@ -571,6 +571,7 @@ ip6_icmp_error (vlib_main_t * vm,
b->current_length = 0;
}
}
+ p0->current_length = p0->current_length > 1280 ? 1280 : p0->current_length;
/* Add IP header and ICMPv6 header including a 4 byte data field */
vlib_buffer_advance(p0,
@@ -581,8 +582,8 @@ ip6_icmp_error (vlib_main_t * vm,
/* Fill ip header fields */
out_ip0->ip_version_traffic_class_and_flow_label =
clib_host_to_net_u32(0x6<<28);
- u16 plen = p0->current_length > 1280 ? 1280 : p0->current_length;
- out_ip0->payload_length = clib_host_to_net_u16(plen - sizeof(ip6_header_t));
+
+ out_ip0->payload_length = clib_host_to_net_u16(p0->current_length - sizeof(ip6_header_t));
out_ip0->protocol = IP_PROTOCOL_ICMP6;
out_ip0->hop_limit = 0xff;
out_ip0->dst_address = ip0->src_address;
diff --git a/vnet/vnet/ip/ip.h b/vnet/vnet/ip/ip.h
index e47512a960d..a0b4ea6d5cb 100644
--- a/vnet/vnet/ip/ip.h
+++ b/vnet/vnet/ip/ip.h
@@ -56,6 +56,7 @@
#include <vnet/ip/ip4.h>
#include <vnet/ip/ip4_error.h>
#include <vnet/ip/ip4_packet.h>
+#include <vnet/ip/icmp4.h>
#include <vnet/ip/ip6.h>
#include <vnet/ip/ip6_packet.h>
diff --git a/vnet/vnet/ip/ip4_input.c b/vnet/vnet/ip/ip4_input.c
index 68edc0fa918..f31df0f6eb9 100644
--- a/vnet/vnet/ip/ip4_input.c
+++ b/vnet/vnet/ip/ip4_input.c
@@ -64,7 +64,7 @@ typedef enum {
IP4_INPUT_NEXT_PUNT,
IP4_INPUT_NEXT_LOOKUP,
IP4_INPUT_NEXT_LOOKUP_MULTICAST,
- IP4_INPUT_NEXT_TTL_EXPIRE,
+ IP4_INPUT_NEXT_ICMP_ERROR,
IP4_INPUT_N_NEXT,
} ip4_input_next_t;
@@ -220,19 +220,21 @@ ip4_input_inline (vlib_main_t * vm,
if (PREDICT_FALSE(error0 != IP4_ERROR_NONE))
{
- next0 = (error0 != IP4_ERROR_OPTIONS
- ? (error0 == IP4_ERROR_TIME_EXPIRED
- ? IP4_INPUT_NEXT_TTL_EXPIRE
- : IP4_INPUT_NEXT_DROP)
- : IP4_INPUT_NEXT_PUNT);
+ if (error0 == IP4_ERROR_TIME_EXPIRED) {
+ icmp4_error_set_vnet_buffer(p0, ICMP4_time_exceeded,
+ ICMP4_time_exceeded_ttl_exceeded_in_transit, 0);
+ next0 = IP4_INPUT_NEXT_ICMP_ERROR;
+ } else
+ next0 = error0 != IP4_ERROR_OPTIONS ? IP4_INPUT_NEXT_DROP : IP4_INPUT_NEXT_PUNT;
}
if (PREDICT_FALSE(error1 != IP4_ERROR_NONE))
{
- next1 = (error1 != IP4_ERROR_OPTIONS
- ? (error1 == IP4_ERROR_TIME_EXPIRED
- ? IP4_INPUT_NEXT_TTL_EXPIRE
- : IP4_INPUT_NEXT_DROP)
- : IP4_INPUT_NEXT_PUNT);
+ if (error1 == IP4_ERROR_TIME_EXPIRED) {
+ icmp4_error_set_vnet_buffer(p1, ICMP4_time_exceeded,
+ ICMP4_time_exceeded_ttl_exceeded_in_transit, 0);
+ next1 = IP4_INPUT_NEXT_ICMP_ERROR;
+ } else
+ next1 = error1 != IP4_ERROR_OPTIONS ? IP4_INPUT_NEXT_DROP : IP4_INPUT_NEXT_PUNT;
}
vlib_validate_buffer_enqueue_x2 (vm, node, next_index,
@@ -307,11 +309,12 @@ ip4_input_inline (vlib_main_t * vm,
p0->error = error_node->errors[error0];
if (PREDICT_FALSE(error0 != IP4_ERROR_NONE))
{
- next0 = (error0 != IP4_ERROR_OPTIONS
- ? (error0 == IP4_ERROR_TIME_EXPIRED
- ? IP4_INPUT_NEXT_TTL_EXPIRE
- : IP4_INPUT_NEXT_DROP)
- : IP4_INPUT_NEXT_PUNT);
+ if (error0 == IP4_ERROR_TIME_EXPIRED) {
+ icmp4_error_set_vnet_buffer(p0, ICMP4_time_exceeded,
+ ICMP4_time_exceeded_ttl_exceeded_in_transit, 0);
+ next0 = IP4_INPUT_NEXT_ICMP_ERROR;
+ } else
+ next0 = error0 != IP4_ERROR_OPTIONS ? IP4_INPUT_NEXT_DROP : IP4_INPUT_NEXT_PUNT;
}
vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
@@ -361,7 +364,7 @@ VLIB_REGISTER_NODE (ip4_input_node) = {
[IP4_INPUT_NEXT_PUNT] = "error-punt",
[IP4_INPUT_NEXT_LOOKUP] = "ip4-lookup",
[IP4_INPUT_NEXT_LOOKUP_MULTICAST] = "ip4-lookup-multicast",
- [IP4_INPUT_NEXT_TTL_EXPIRE] = "ip4-icmp-ttl-expire",
+ [IP4_INPUT_NEXT_ICMP_ERROR] = "ip4-icmp-error",
},
.format_buffer = format_ip4_header,
@@ -379,7 +382,7 @@ VLIB_REGISTER_NODE (ip4_input_no_checksum_node,static) = {
[IP4_INPUT_NEXT_PUNT] = "error-punt",
[IP4_INPUT_NEXT_LOOKUP] = "ip4-lookup",
[IP4_INPUT_NEXT_LOOKUP_MULTICAST] = "ip4-lookup-multicast",
- [IP4_INPUT_NEXT_TTL_EXPIRE] = "ip4-icmp-ttl-expire",
+ [IP4_INPUT_NEXT_ICMP_ERROR] = "ip4-icmp-error",
},
.format_buffer = format_ip4_header,
diff --git a/vnet/vnet/ip/ip6_input.c b/vnet/vnet/ip/ip6_input.c
index 473b2b2719f..f96a1cfb5a5 100644
--- a/vnet/vnet/ip/ip6_input.c
+++ b/vnet/vnet/ip/ip6_input.c
@@ -62,7 +62,7 @@ static u8 * format_ip6_input_trace (u8 * s, va_list * va)
typedef enum {
IP6_INPUT_NEXT_DROP,
IP6_INPUT_NEXT_LOOKUP,
- IP6_INPUT_NEXT_ICMP,
+ IP6_INPUT_NEXT_ICMP_ERROR,
IP6_INPUT_N_NEXT,
} ip6_input_next_t;
@@ -189,7 +189,7 @@ ip6_input (vlib_main_t * vm,
if (error0 == IP6_ERROR_TIME_EXPIRED) {
icmp6_error_set_vnet_buffer(p0, ICMP6_time_exceeded,
ICMP6_time_exceeded_ttl_exceeded_in_transit, 0);
- next0 = IP6_INPUT_NEXT_ICMP;
+ next0 = IP6_INPUT_NEXT_ICMP_ERROR;
} else {
next0 = IP6_INPUT_NEXT_DROP;
}
@@ -199,7 +199,7 @@ ip6_input (vlib_main_t * vm,
if (error1 == IP6_ERROR_TIME_EXPIRED) {
icmp6_error_set_vnet_buffer(p1, ICMP6_time_exceeded,
ICMP6_time_exceeded_ttl_exceeded_in_transit, 0);
- next1 = IP6_INPUT_NEXT_ICMP;
+ next1 = IP6_INPUT_NEXT_ICMP_ERROR;
} else {
next1 = IP6_INPUT_NEXT_DROP;
}
@@ -262,7 +262,7 @@ ip6_input (vlib_main_t * vm,
if (error0 == IP6_ERROR_TIME_EXPIRED) {
icmp6_error_set_vnet_buffer(p0, ICMP6_time_exceeded,
ICMP6_time_exceeded_ttl_exceeded_in_transit, 0);
- next0 = IP6_INPUT_NEXT_ICMP;
+ next0 = IP6_INPUT_NEXT_ICMP_ERROR;
} else {
next0 = IP6_INPUT_NEXT_DROP;
}
@@ -298,7 +298,7 @@ VLIB_REGISTER_NODE (ip6_input_node) = {
.next_nodes = {
[IP6_INPUT_NEXT_DROP] = "error-drop",
[IP6_INPUT_NEXT_LOOKUP] = "ip6-lookup",
- [IP6_INPUT_NEXT_ICMP] = "ip6-icmp-error",
+ [IP6_INPUT_NEXT_ICMP_ERROR] = "ip6-icmp-error",
},
.format_buffer = format_ip6_header,
diff --git a/vnet/vnet/ip/ip_frag.c b/vnet/vnet/ip/ip_frag.c
index 22176187a9c..3436090a426 100644
--- a/vnet/vnet/ip/ip_frag.c
+++ b/vnet/vnet/ip/ip_frag.c
@@ -37,9 +37,8 @@ static u8 * format_ip_frag_trace (u8 * s, va_list * args)
CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
ip_frag_trace_t * t = va_arg (*args, ip_frag_trace_t *);
- s = format(s, "IPv%s offset: %u mtu: %u fragments: %u next: %s",
- t->ipv6?"6":"4",
- t->header_offset, t->mtu, t->n_fragments, node->next_node_names[t->next]);
+ s = format(s, "IPv%s offset: %u mtu: %u fragments: %u",
+ t->ipv6?"6":"4", t->header_offset, t->mtu, t->n_fragments);
return s;
}
@@ -146,6 +145,14 @@ ip4_frag_do_fragment(vlib_main_t *vm, u32 pi, u32 **buffer, ip_frag_error_t *err
}
}
+void
+ip_frag_set_vnet_buffer (vlib_buffer_t *b, u16 offset, u16 mtu, u8 next_index, u8 flags)
+{
+ vnet_buffer(b)->ip_frag.header_offset = offset;
+ vnet_buffer(b)->ip_frag.mtu = mtu;
+ vnet_buffer(b)->ip_frag.next_index = next_index;
+ vnet_buffer(b)->ip_frag.flags = flags;
+}
static uword
ip4_frag (vlib_main_t *vm,
@@ -189,13 +196,25 @@ ip4_frag (vlib_main_t *vm,
tr->next = vnet_buffer(p0)->ip_frag.next_index;
}
- next0 = (error0 == IP_FRAG_ERROR_NONE) ? vnet_buffer(p0)->ip_frag.next_index : IP4_FRAG_NEXT_DROP;
- frag_sent += vec_len(buffer);
- small_packets += (vec_len(buffer) == 1);
+ if (error0 == IP_FRAG_ERROR_DONT_FRAGMENT_SET) {
+ icmp4_error_set_vnet_buffer(p0, ICMP4_destination_unreachable,
+ ICMP4_destination_unreachable_fragmentation_needed_and_dont_fragment_set,
+ vnet_buffer(p0)->ip_frag.mtu);
+ vlib_buffer_advance(p0, vnet_buffer(p0)->ip_frag.header_offset);
+ next0 = IP4_FRAG_NEXT_ICMP_ERROR;
+ } else
+ next0 = (error0 == IP_FRAG_ERROR_NONE) ? vnet_buffer(p0)->ip_frag.next_index : IP4_FRAG_NEXT_DROP;
+
+ if (error0 == IP_FRAG_ERROR_NONE) {
+ frag_sent += vec_len(buffer);
+ small_packets += (vec_len(buffer) == 1);
+ } else
+ vlib_error_count(vm, ip4_frag_node.index, error0, 1);
//Send fragments that were added in the frame
frag_from = buffer;
frag_left = vec_len(buffer);
+
while (frag_left > 0) {
while (frag_left > 0 && n_left_to_next > 0) {
u32 i;
@@ -218,6 +237,7 @@ ip4_frag (vlib_main_t *vm,
vlib_put_next_frame(vm, node, next_index, n_left_to_next);
}
vec_free(buffer);
+
vlib_node_increment_counter(vm, ip4_frag_node.index, IP_FRAG_ERROR_FRAGMENT_SENT, frag_sent);
vlib_node_increment_counter(vm, ip4_frag_node.index, IP_FRAG_ERROR_SMALL_PACKET, small_packets);
@@ -426,6 +446,7 @@ VLIB_REGISTER_NODE (ip4_frag_node) = {
.next_nodes = {
[IP4_FRAG_NEXT_IP4_LOOKUP] = "ip4-lookup",
[IP4_FRAG_NEXT_IP6_LOOKUP] = "ip6-lookup",
+ [IP4_FRAG_NEXT_ICMP_ERROR] = "ip4-icmp-error",
[IP4_FRAG_NEXT_DROP] = "error-drop"
},
};
diff --git a/vnet/vnet/ip/ip_frag.h b/vnet/vnet/ip/ip_frag.h
index 04566904e5f..76238655af9 100644
--- a/vnet/vnet/ip/ip_frag.h
+++ b/vnet/vnet/ip/ip_frag.h
@@ -49,6 +49,7 @@ vlib_node_registration_t ip6_frag_node;
typedef enum {
IP4_FRAG_NEXT_IP4_LOOKUP,
IP4_FRAG_NEXT_IP6_LOOKUP,
+ IP4_FRAG_NEXT_ICMP_ERROR,
IP4_FRAG_NEXT_DROP,
IP4_FRAG_N_NEXT
} ip4_frag_next_t;
@@ -65,8 +66,8 @@ typedef enum {
_(NONE, "packet fragmented") \
_(SMALL_PACKET, "packet smaller than MTU") \
_(FRAGMENT_SENT, "number of sent fragments") \
- _(CANT_FRAGMENT_HEADER, "can't fragment header'") \
- _(DONT_FRAGMENT_SET, "can't fragment this packet'") \
+ _(CANT_FRAGMENT_HEADER, "can't fragment header") \
+ _(DONT_FRAGMENT_SET, "can't fragment this packet") \
_(MALFORMED, "malformed packet") \
_(MEMORY, "could not allocate buffer") \
_(UNKNOWN, "unknown error")
@@ -78,4 +79,6 @@ typedef enum {
IP_FRAG_N_ERROR,
} ip_frag_error_t;
+void ip_frag_set_vnet_buffer(vlib_buffer_t *b, u16 offset, u16 mtu, u8 next_index, u8 flags);
+
#endif /* ifndef IP_FRAG_H */
diff --git a/vnet/vnet/map/ip4_map.c b/vnet/vnet/map/ip4_map.c
index 343b57d210f..7b9b3ed9d89 100644
--- a/vnet/vnet/map/ip4_map.c
+++ b/vnet/vnet/map/ip4_map.c
@@ -27,8 +27,10 @@ enum ip4_map_next_e {
#ifdef MAP_SKIP_IP6_LOOKUP
IP4_MAP_NEXT_IP6_REWRITE,
#endif
- IP4_MAP_NEXT_FRAGMENT,
+ IP4_MAP_NEXT_IP4_FRAGMENT,
+ IP4_MAP_NEXT_IP6_FRAGMENT,
IP4_MAP_NEXT_REASS,
+ IP4_MAP_NEXT_ICMP_ERROR,
IP4_MAP_NEXT_DROP,
IP4_MAP_N_NEXT,
};
@@ -177,6 +179,27 @@ ip4_map_decrement_ttl (ip4_header_t *ip, u8 *error)
ASSERT (ip->checksum == ip4_header_checksum(ip));
}
+static u32
+ip4_map_fragment (vlib_buffer_t *b, u16 mtu, bool df, u8 *error)
+{
+ map_main_t *mm = &map_main;
+
+ if (mm->frag_inner) {
+ ip_frag_set_vnet_buffer(b, sizeof(ip6_header_t), mtu, IP4_FRAG_NEXT_IP6_LOOKUP, IP_FRAG_FLAG_IP6_HEADER);
+ return (IP4_MAP_NEXT_IP4_FRAGMENT);
+ } else {
+ if (df && !mm->frag_ignore_df) {
+ icmp4_error_set_vnet_buffer(b, ICMP4_destination_unreachable,
+ ICMP4_destination_unreachable_fragmentation_needed_and_dont_fragment_set, mtu);
+ vlib_buffer_advance(b, sizeof(ip6_header_t));
+ *error = MAP_ERROR_DF_SET;
+ return (IP4_MAP_NEXT_ICMP_ERROR);
+ }
+ ip_frag_set_vnet_buffer(b, 0, mtu, IP6_FRAG_NEXT_IP6_LOOKUP, IP_FRAG_FLAG_IP6_HEADER);
+ return (IP4_MAP_NEXT_IP6_FRAGMENT);
+ }
+}
+
/*
* ip4_map
*/
@@ -247,6 +270,12 @@ ip4_map (vlib_main_t *vm,
port0 = ip4_map_port_and_security_check(d0, ip40, &next0, &error0);
port1 = ip4_map_port_and_security_check(d1, ip41, &next1, &error1);
+ /* Decrement IPv4 TTL */
+ ip4_map_decrement_ttl(ip40, &error0);
+ ip4_map_decrement_ttl(ip41, &error1);
+ bool df0 = ip40->flags_and_fragment_offset & clib_host_to_net_u16(IP4_HEADER_FLAG_DONT_FRAGMENT);
+ bool df1 = ip41->flags_and_fragment_offset & clib_host_to_net_u16(IP4_HEADER_FLAG_DONT_FRAGMENT);
+
/* MAP calc */
u32 da40 = clib_net_to_host_u32(ip40->dst_address.as_u32);
u32 da41 = clib_net_to_host_u32(ip41->dst_address.as_u32);
@@ -288,11 +317,7 @@ ip4_map (vlib_main_t *vm,
*/
if (PREDICT_TRUE(error0 == MAP_ERROR_NONE)) {
if (PREDICT_FALSE(d0->mtu && (clib_net_to_host_u16(ip6h0->payload_length) + sizeof(*ip6h0) > d0->mtu))) {
- vnet_buffer(p0)->ip_frag.header_offset = sizeof(*ip6h0);
- vnet_buffer(p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP6_LOOKUP;
- vnet_buffer(p0)->ip_frag.mtu = d0->mtu;
- vnet_buffer(p0)->ip_frag.flags = IP_FRAG_FLAG_IP6_HEADER;
- next0 = IP4_MAP_NEXT_FRAGMENT;
+ next0 = ip4_map_fragment(p0, d0->mtu, df0, &error0);
} else {
next0 = ip4_map_ip6_lookup_bypass(p0, ip40) ? IP4_MAP_NEXT_IP6_REWRITE : next0;
vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, map_domain_index0, 1,
@@ -308,11 +333,7 @@ ip4_map (vlib_main_t *vm,
*/
if (PREDICT_TRUE(error1 == MAP_ERROR_NONE)) {
if (PREDICT_FALSE(d1->mtu && (clib_net_to_host_u16(ip6h1->payload_length) + sizeof(*ip6h1) > d1->mtu))) {
- vnet_buffer(p1)->ip_frag.header_offset = sizeof(*ip6h1);
- vnet_buffer(p1)->ip_frag.next_index = IP4_FRAG_NEXT_IP6_LOOKUP;
- vnet_buffer(p1)->ip_frag.mtu = d1->mtu;
- vnet_buffer(p1)->ip_frag.flags = IP_FRAG_FLAG_IP6_HEADER;
- next1 = IP4_MAP_NEXT_FRAGMENT;
+ next1 = ip4_map_fragment(p1, d1->mtu, df1, &error1);
} else {
next1 = ip4_map_ip6_lookup_bypass(p1, ip41) ? IP4_MAP_NEXT_IP6_REWRITE : next1;
vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, map_domain_index1, 1,
@@ -369,6 +390,7 @@ ip4_map (vlib_main_t *vm,
/* Decrement IPv4 TTL */
ip4_map_decrement_ttl(ip40, &error0);
+ bool df0 = ip40->flags_and_fragment_offset & clib_host_to_net_u16(IP4_HEADER_FLAG_DONT_FRAGMENT);
/* MAP calc */
u32 da40 = clib_net_to_host_u32(ip40->dst_address.as_u32);
@@ -396,11 +418,7 @@ ip4_map (vlib_main_t *vm,
*/
if (PREDICT_TRUE(error0 == MAP_ERROR_NONE)) {
if (PREDICT_FALSE(d0->mtu && (clib_net_to_host_u16(ip6h0->payload_length) + sizeof(*ip6h0) > d0->mtu))) {
- vnet_buffer(p0)->ip_frag.header_offset = sizeof(*ip6h0);
- vnet_buffer(p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP6_LOOKUP;
- vnet_buffer(p0)->ip_frag.mtu = d0->mtu;
- vnet_buffer(p0)->ip_frag.flags = IP_FRAG_FLAG_IP6_HEADER;
- next0 = IP4_MAP_NEXT_FRAGMENT;
+ next0 = ip4_map_fragment(p0, d0->mtu, df0, &error0);
} else {
next0 = ip4_map_ip6_lookup_bypass(p0, ip40) ? IP4_MAP_NEXT_IP6_REWRITE : next0;
vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, map_domain_index0, 1,
@@ -591,8 +609,10 @@ VLIB_REGISTER_NODE(ip4_map_node) = {
#ifdef MAP_SKIP_IP6_LOOKUP
[IP4_MAP_NEXT_IP6_REWRITE] = "ip6-rewrite",
#endif
- [IP4_MAP_NEXT_FRAGMENT] = "ip4-frag",
+ [IP4_MAP_NEXT_IP4_FRAGMENT] = "ip4-frag",
+ [IP4_MAP_NEXT_IP6_FRAGMENT] = "ip6-frag",
[IP4_MAP_NEXT_REASS] = "ip4-map-reass",
+ [IP4_MAP_NEXT_ICMP_ERROR] = "ip4-icmp-error",
[IP4_MAP_NEXT_DROP] = "error-drop",
},
};
diff --git a/vnet/vnet/map/map.c b/vnet/vnet/map/map.c
index eb5496b8829..a63122b58f3 100644
--- a/vnet/vnet/map/map.c
+++ b/vnet/vnet/map/map.c
@@ -653,6 +653,58 @@ map_icmp_unreachables_command_fn (vlib_main_t *vm,
}
static clib_error_t *
+map_fragment_command_fn (vlib_main_t *vm,
+ unformat_input_t *input,
+ vlib_cli_command_t *cmd)
+{
+ unformat_input_t _line_input, *line_input = &_line_input;
+ map_main_t *mm = &map_main;
+
+ /* Get a line of input. */
+ if (!unformat_user(input, unformat_line_input, line_input))
+ return 0;
+
+ while (unformat_check_input(line_input) != UNFORMAT_END_OF_INPUT) {
+ if (unformat(line_input, "inner"))
+ mm->frag_inner = true;
+ else if (unformat(line_input, "outer"))
+ mm->frag_inner = false;
+ else
+ return clib_error_return(0, "unknown input `%U'",
+ format_unformat_error, input);
+ }
+ unformat_free(line_input);
+
+ return 0;
+}
+
+static clib_error_t *
+map_fragment_df_command_fn (vlib_main_t *vm,
+ unformat_input_t *input,
+ vlib_cli_command_t *cmd)
+{
+ unformat_input_t _line_input, *line_input = &_line_input;
+ map_main_t *mm = &map_main;
+
+ /* Get a line of input. */
+ if (!unformat_user(input, unformat_line_input, line_input))
+ return 0;
+
+ while (unformat_check_input(line_input) != UNFORMAT_END_OF_INPUT) {
+ if (unformat(line_input, "on"))
+ mm->frag_ignore_df = true;
+ else if (unformat(line_input, "off"))
+ mm->frag_ignore_df = false;
+ else
+ return clib_error_return(0, "unknown input `%U'",
+ format_unformat_error, input);
+ }
+ unformat_free(line_input);
+
+ return 0;
+}
+
+static clib_error_t *
map_traffic_class_command_fn (vlib_main_t *vm,
unformat_input_t *input,
vlib_cli_command_t *cmd)
@@ -869,6 +921,8 @@ show_map_stats_command_fn (vlib_main_t *vm, unformat_input_t *input, vlib_cli_co
vlib_cli_output(vm, "ICMP-relay IPv4 source address: %U\n", format_ip4_address, &mm->icmp4_src_address);
vlib_cli_output(vm, "ICMP6 unreachables sent for unmatched packets: %s\n", mm->icmp6_enabled ? "enabled" : "disabled");
+ vlib_cli_output(vm, "Inner fragmentation: %s\n", mm->frag_inner ? "enabled" : "disabled");
+ vlib_cli_output(vm, "Fragment packets regardless of DF flag: %s\n", mm->frag_ignore_df ? "enabled" : "disabled");
/*
* Counters
@@ -1563,11 +1617,23 @@ VLIB_CLI_COMMAND(map_icmp_relay_source_address_command, static) = {
};
VLIB_CLI_COMMAND(map_icmp_unreachables_command, static) = {
- .path = "map params icmp unreachables",
+ .path = "map params icmp6 unreachables",
.short_help = "unreachables {on|off}",
.function = map_icmp_unreachables_command_fn,
};
+VLIB_CLI_COMMAND(map_fragment_command, static) = {
+ .path = "map params fragment",
+ .short_help = "[inner|outer] [ignore-df [on|off]]",
+ .function = map_fragment_command_fn,
+};
+
+VLIB_CLI_COMMAND(map_fragment_df_command, static) = {
+ .path = "map params fragment ignore-df",
+ .short_help = "on|off",
+ .function = map_fragment_df_command_fn,
+};
+
VLIB_CLI_COMMAND(map_security_check_frag_command, static) = {
.path = "map params security-check fragments",
.short_help =
@@ -1639,6 +1705,10 @@ clib_error_t *map_init (vlib_main_t *vm)
/* ICMP6 Type 1, Code 5 for security check failure */
mm->icmp6_enabled = false;
+ /* Inner or outer fragmentation */
+ mm->frag_inner = false;
+ mm->frag_ignore_df = false;
+
vec_validate(mm->domain_counters, MAP_N_DOMAIN_COUNTER - 1);
mm->domain_counters[MAP_DOMAIN_COUNTER_RX].name = "rx";
mm->domain_counters[MAP_DOMAIN_COUNTER_TX].name = "tx";
diff --git a/vnet/vnet/map/map.h b/vnet/vnet/map/map.h
index 6d12b71be2b..d38d7f47241 100644
--- a/vnet/vnet/map/map.h
+++ b/vnet/vnet/map/map.h
@@ -190,9 +190,6 @@ typedef struct {
vlib_combined_counter_main_t *domain_counters;
volatile u32 *counter_lock;
- /* Global counters */
- vlib_simple_counter_main_t icmp_relayed;
-
#ifdef MAP_SKIP_IP6_LOOKUP
/* pre-presolve */
u32 adj6_index, adj4_index;
@@ -203,12 +200,14 @@ typedef struct {
/* Traffic class: zero, copy (~0) or fixed value */
u8 tc;
bool tc_copy;
- bool sec_check;
- bool sec_check_frag;
- bool icmp6_enabled;
+
+ bool sec_check; /* Inbound security check */
+ bool sec_check_frag; /* Inbound security check for (subsequent) fragments */
+ bool icmp6_enabled; /* Send destination unreachable for security check failure */
/* ICMPv6 -> ICMPv4 relay parameters */
ip4_address_t icmp4_src_address;
+ vlib_simple_counter_main_t icmp_relayed;
/* convenience */
vlib_main_t *vlib_main;
@@ -217,13 +216,13 @@ typedef struct {
/*
* IPv4 encap and decap reassembly
*/
- //Conf
+ /* Configuration */
f32 ip4_reass_conf_ht_ratio; //Size of ht is 2^ceil(log2(ratio*pool_size))
u16 ip4_reass_conf_pool_size; //Max number of allocated reass structures
u16 ip4_reass_conf_lifetime_ms; //Time a reassembly struct is considered valid in ms
u32 ip4_reass_conf_buffers; //Maximum number of buffers used by ip4 reassembly
- //Runtime
+ /* Runtime */
map_ip4_reass_t *ip4_reass_pool;
u8 ip4_reass_ht_log2len; //Hash table size is 2^log2len
u16 ip4_reass_allocated;
@@ -231,19 +230,22 @@ typedef struct {
u16 ip4_reass_fifo_last;
volatile u32 *ip4_reass_lock;
- //Counters
+ /* Counters */
u32 ip4_reass_buffered_counter;
+ bool frag_inner; /* Inner or outer fragmentation */
+ bool frag_ignore_df; /* Fragment (outer) packet even if DF is set */
+
/*
* IPv6 decap reassembly
*/
- //Conf
+ /* Configuration */
f32 ip6_reass_conf_ht_ratio; //Size of ht is 2^ceil(log2(ratio*pool_size))
u16 ip6_reass_conf_pool_size; //Max number of allocated reass structures
u16 ip6_reass_conf_lifetime_ms; //Time a reassembly struct is considered valid in ms
u32 ip6_reass_conf_buffers; //Maximum number of buffers used by ip6 reassembly
- //Runtime
+ /* Runtime */
map_ip6_reass_t *ip6_reass_pool;
u8 ip6_reass_ht_log2len; //Hash table size is 2^log2len
u16 ip6_reass_allocated;
@@ -251,19 +253,18 @@ typedef struct {
u16 ip6_reass_fifo_last;
volatile u32 *ip6_reass_lock;
- //Counters
+ /* Counters */
u32 ip6_reass_buffered_counter;
} map_main_t;
/*
- * TODO: Remove SEC_CHECK / TRANSLATED_4TO6 / TRANSLATED_6TO4
+ * MAP Error counters/messages
*/
#define foreach_map_error \
/* Must be first. */ \
_(NONE, "valid MAP packets") \
_(BAD_PROTOCOL, "bad protocol") \
- _(WRONG_ICMP_TYPE, "wrong icmp type") \
_(SEC_CHECK, "security check failed") \
_(ENCAP_SEC_CHECK, "encap security check failed") \
_(DECAP_SEC_CHECK, "decap security check failed") \
@@ -277,7 +278,7 @@ typedef struct {
_(FRAGMENT_MALFORMED, "fragment has unexpected format")\
_(FRAGMENT_DROPPED, "dropped cached fragment") \
_(MALFORMED, "malformed packet") \
- _(IP4_ERROR_TIME_EXPIRED, "time expired")
+ _(DF_SET, "can't fragment, DF set")
typedef enum {
#define _(sym,str) MAP_ERROR_##sym,