[ovs-dev] OVN: add ICMP time exceeded support to OVN logical router

Message ID a2ce3f28fd88aaa09ab28ef3edd9234b6f364d4a.1528214013.git.lorenzo.bianconi@redhat.com
State Superseded
Headers show
Series
  • [ovs-dev] OVN: add ICMP time exceeded support to OVN logical router
Related show

Commit Message

Lorenzo Bianconi June 5, 2018, 3:57 p.m.
Using icmp4 action, send an ICMP time exceeded frame whenever
an OVN logical router receives an IPv4 packets whose TTL has
expired (ip.ttl == {0, 1})

Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@redhat.com>
---
 ovn/northd/ovn-northd.8.xml |  4 ---
 ovn/northd/ovn-northd.c     | 26 +++++++++++++--
 tests/ovn.at                | 77 +++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 100 insertions(+), 7 deletions(-)

Comments

Mark Michelson June 13, 2018, 7:23 p.m. | #1
Hi Lorenzo, the patch looks good to me. I have one comment on the test 
you added, though. See below.

On 06/05/2018 11:57 AM, Lorenzo Bianconi wrote:
> Using icmp4 action, send an ICMP time exceeded frame whenever
> an OVN logical router receives an IPv4 packets whose TTL has
> expired (ip.ttl == {0, 1})
> 
> Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@redhat.com>
> ---
>   ovn/northd/ovn-northd.8.xml |  4 ---
>   ovn/northd/ovn-northd.c     | 26 +++++++++++++--
>   tests/ovn.at                | 77 +++++++++++++++++++++++++++++++++++++++++++++
>   3 files changed, 100 insertions(+), 7 deletions(-)
> 
> diff --git a/ovn/northd/ovn-northd.8.xml b/ovn/northd/ovn-northd.8.xml
> index 1d68f1aab..759d3dace 100644
> --- a/ovn/northd/ovn-northd.8.xml
> +++ b/ovn/northd/ovn-northd.8.xml
> @@ -1401,10 +1401,6 @@ icmp4 {
>       next;
>   };
>           </pre>
> -
> -        <p>
> -          Not yet implemented.
> -        </p>
>         </li>
>   
>         <li>
> diff --git a/ovn/northd/ovn-northd.c b/ovn/northd/ovn-northd.c
> index 0e06776ad..60f43e2a6 100644
> --- a/ovn/northd/ovn-northd.c
> +++ b/ovn/northd/ovn-northd.c
> @@ -4852,9 +4852,7 @@ build_lrouter_flows(struct hmap *datapaths, struct hmap *ports,
>           ovn_lflow_add(lflows, od, S_ROUTER_IN_IP_INPUT, 50,
>                         "eth.bcast", "drop;");
>   
> -        /* TTL discard.
> -         *
> -         * XXX Need to send ICMP time exceeded if !ip.later_frag. */
> +        /* TTL discard */
>           ds_clear(&match);
>           ds_put_cstr(&match, "ip4 && ip.ttl == {0, 1}");
>           ovn_lflow_add(lflows, od, S_ROUTER_IN_IP_INPUT, 30,
> @@ -4920,6 +4918,28 @@ build_lrouter_flows(struct hmap *datapaths, struct hmap *ports,
>                             ds_cstr(&match), ds_cstr(&actions));
>           }
>   
> +        /* ICMP time exceeded */
> +        for (int i = 0; i < op->lrp_networks.n_ipv4_addrs; i++) {
> +            ds_clear(&match);
> +            ds_clear(&actions);
> +
> +            ds_put_format(&match,
> +                          "inport == %s && ip4 && "
> +                          "ip.ttl == {0, 1} && !ip.later_frag", op->json_key);
> +            ds_put_format(&actions,
> +                          "icmp4 {"
> +                          "eth.dst <-> eth.src; "
> +                          "icmp4.type = 11; /* Time exceeded */ "
> +                          "icmp4.code = 0; /* TTL exceeded in transit */ "
> +                          "ip4.dst = ip4.src; "
> +                          "ip4.src = %s; "
> +                          "ip.ttl = 255; "
> +                          "next; };",
> +                          op->lrp_networks.ipv4_addrs[i].addr_s);
> +            ovn_lflow_add(lflows, op->od, S_ROUTER_IN_IP_INPUT, 40,
> +                          ds_cstr(&match), ds_cstr(&actions));
> +        }
> +
>           /* ARP reply.  These flows reply to ARP requests for the router's own
>            * IP address. */
>           for (int i = 0; i < op->lrp_networks.n_ipv4_addrs; i++) {
> diff --git a/tests/ovn.at b/tests/ovn.at
> index f12c24c17..81c834276 100644
> --- a/tests/ovn.at
> +++ b/tests/ovn.at
> @@ -10267,3 +10267,80 @@ $PYTHON "$top_srcdir/utilities/ovs-pcap.in" hv1/vif2-tx.pcap > 2.packets
>   AT_CHECK([cat 2.packets], [0], [])
>   
>   AT_CLEANUP
> +
> +AT_SETUP([ovn -- TTL exceeded])
> +AT_KEYWORDS([ttl-exceeded])
> +AT_SKIP_IF([test $HAVE_PYTHON = no])
> +ovn_start
> +
> +# test_ip_packet INPORT HV ETH_SRC ETH_DST IPV4_SRC IPV4_DST IPV4_ROUTER IP_CHKSUM EXP_IP_CHKSUM EXP_ICMP_CHKSUM
> +#
> +# Causes a packet to be received on INPORT of the hypervisor HV. The packet is an IPv4 packet with
> +# ETH_SRC, ETH_DST, IPV4_SRC, IPV4_DST, IP_CHKSUM as specified and TTL set to 1.
> +# EXP_IP_CHKSUM and EXP_ICMP_CHKSUM are the ip and icmp checksums of the icmp time exceeded frame
> +# generated by OVN logical router
> +#
> +# INPORT is a lport number, e.g. 11 for vif11.
> +# HV is a hypervisor number
> +# ETH_SRC and ETH_DST are each 12 hex digits.
> +# IPV4_SRC, IPV4_DST and IPV4_ROUTER are each 8 hex digits.
> +# IP_CHKSUM, EXP_IP_CHSUM and EXP_ICMP_CHKSUM are each 4 hex digits
> +test_ip_packet() {
> +    local inport=$1 hv=$2 eth_src=$3 eth_dst=$4 ipv4_src=$5 ipv4_dst=$6 ip_router=$7 ip_chksum=$8
> +    local exp_ip_chksum=$9 exp_icmp_chksum=${10}
> +    shift 10
> +
> +    local ip_ttl=01
> +    local packet=${eth_dst}${eth_src}08004500001400004000${ip_ttl}01${ip_chksum}${ipv4_src}${ipv4_dst}
> +
> +    local reply_icmp_ttl=fe
> +    local icmp_type_code_response=0b00
> +    local icmp_data=00000000
> +    local reply_icmp_payload=${icmp_type_code_response}${exp_icmp_chksum}${icmp_data}
> +    local reply=${eth_src}${eth_dst}08004500001c00004000${reply_icmp_ttl}01${exp_ip_chksum}${ip_router}${ipv4_src}${reply_icmp_payload}
> +    echo $reply >> vif$inport.expected
> +
> +    as hv$hv ovs-appctl netdev-dummy/receive vif$inport $packet
> +}
> +
> +ip_to_hex() {
> +    printf "%02x%02x%02x%02x" "$@"
> +}
> +
> +for i in 1 2; do
> +    net_add n$i
> +    ovn-nbctl ls-add sw$i
> +
> +    sim_add hv$i
> +    as hv$i
> +    ovs-vsctl add-br br-phys
> +    ovn_attach n$i br-phys 192.168.$i.1
> +
> +    ovn-nbctl lsp-add sw$i sw$i-p${i}0 -- \
> +        lsp-set-addresses sw$i-p${i}0 "00:00:00:00:00:0$i 192.168.$i.1"
> +
> +    ovs-vsctl -- add-port br-int vif$i -- \
> +        set interface vif$i \
> +        external-ids:iface-id=sw$i-p${i}0 \
> +            options:tx_pcap=hv$i/vif$i-tx.pcap \
> +            options:rxq_pcap=hv$i/vif$i-rx.pcap \
> +            ofport-request=$i
> +done
> +
> +ovn-nbctl lr-add lr0
> +for i in 1 2; do
> +    ovn-nbctl lrp-add lr0 lrp$i 00:00:00:00:ff:0$i 192.168.$i.254/24
> +    ovn-nbctl -- lsp-add sw$i lrp$i-attachment \
> +              -- set Logical_Switch_Port lrp$i-attachment type=router \
> +                options:router-port=lrp$i addresses='"00:00:00:00:ff:'0$i'"'
> +done
> +
> +OVN_POPULATE_ARP
> +# allow some time for ovn-northd and ovn-controller to catch up.
> +sleep 1

If Han's commits have told me anything, these sleeps are not guaranteed 
to always work. It's probably better to use a proper wait command to 
ensure that everything is synced. You could either add --wait=hv to the 
last ovn-nbctl command above. Or you could change this sleep to 
"ovn-nbctl --wait=hv sync"

> +
> +test_ip_packet 1 1 000000000001 00000000ff01 $(ip_to_hex 192 168 1 1) $(ip_to_hex 192 168 2 1) $(ip_to_hex 192 168 1 254) 0000 7dae f4ff
> +OVN_CHECK_PACKETS([hv1/vif1-tx.pcap], [vif1.expected])
> +
> +OVN_CLEANUP([hv1], [hv2])
> +AT_CLEANUP
>

Patch

diff --git a/ovn/northd/ovn-northd.8.xml b/ovn/northd/ovn-northd.8.xml
index 1d68f1aab..759d3dace 100644
--- a/ovn/northd/ovn-northd.8.xml
+++ b/ovn/northd/ovn-northd.8.xml
@@ -1401,10 +1401,6 @@  icmp4 {
     next;
 };
         </pre>
-
-        <p>
-          Not yet implemented.
-        </p>
       </li>
 
       <li>
diff --git a/ovn/northd/ovn-northd.c b/ovn/northd/ovn-northd.c
index 0e06776ad..60f43e2a6 100644
--- a/ovn/northd/ovn-northd.c
+++ b/ovn/northd/ovn-northd.c
@@ -4852,9 +4852,7 @@  build_lrouter_flows(struct hmap *datapaths, struct hmap *ports,
         ovn_lflow_add(lflows, od, S_ROUTER_IN_IP_INPUT, 50,
                       "eth.bcast", "drop;");
 
-        /* TTL discard.
-         *
-         * XXX Need to send ICMP time exceeded if !ip.later_frag. */
+        /* TTL discard */
         ds_clear(&match);
         ds_put_cstr(&match, "ip4 && ip.ttl == {0, 1}");
         ovn_lflow_add(lflows, od, S_ROUTER_IN_IP_INPUT, 30,
@@ -4920,6 +4918,28 @@  build_lrouter_flows(struct hmap *datapaths, struct hmap *ports,
                           ds_cstr(&match), ds_cstr(&actions));
         }
 
+        /* ICMP time exceeded */
+        for (int i = 0; i < op->lrp_networks.n_ipv4_addrs; i++) {
+            ds_clear(&match);
+            ds_clear(&actions);
+
+            ds_put_format(&match,
+                          "inport == %s && ip4 && "
+                          "ip.ttl == {0, 1} && !ip.later_frag", op->json_key);
+            ds_put_format(&actions,
+                          "icmp4 {"
+                          "eth.dst <-> eth.src; "
+                          "icmp4.type = 11; /* Time exceeded */ "
+                          "icmp4.code = 0; /* TTL exceeded in transit */ "
+                          "ip4.dst = ip4.src; "
+                          "ip4.src = %s; "
+                          "ip.ttl = 255; "
+                          "next; };",
+                          op->lrp_networks.ipv4_addrs[i].addr_s);
+            ovn_lflow_add(lflows, op->od, S_ROUTER_IN_IP_INPUT, 40,
+                          ds_cstr(&match), ds_cstr(&actions));
+        }
+
         /* ARP reply.  These flows reply to ARP requests for the router's own
          * IP address. */
         for (int i = 0; i < op->lrp_networks.n_ipv4_addrs; i++) {
diff --git a/tests/ovn.at b/tests/ovn.at
index f12c24c17..81c834276 100644
--- a/tests/ovn.at
+++ b/tests/ovn.at
@@ -10267,3 +10267,80 @@  $PYTHON "$top_srcdir/utilities/ovs-pcap.in" hv1/vif2-tx.pcap > 2.packets
 AT_CHECK([cat 2.packets], [0], [])
 
 AT_CLEANUP
+
+AT_SETUP([ovn -- TTL exceeded])
+AT_KEYWORDS([ttl-exceeded])
+AT_SKIP_IF([test $HAVE_PYTHON = no])
+ovn_start
+
+# test_ip_packet INPORT HV ETH_SRC ETH_DST IPV4_SRC IPV4_DST IPV4_ROUTER IP_CHKSUM EXP_IP_CHKSUM EXP_ICMP_CHKSUM
+#
+# Causes a packet to be received on INPORT of the hypervisor HV. The packet is an IPv4 packet with
+# ETH_SRC, ETH_DST, IPV4_SRC, IPV4_DST, IP_CHKSUM as specified and TTL set to 1.
+# EXP_IP_CHKSUM and EXP_ICMP_CHKSUM are the ip and icmp checksums of the icmp time exceeded frame
+# generated by OVN logical router
+#
+# INPORT is a lport number, e.g. 11 for vif11.
+# HV is a hypervisor number
+# ETH_SRC and ETH_DST are each 12 hex digits.
+# IPV4_SRC, IPV4_DST and IPV4_ROUTER are each 8 hex digits.
+# IP_CHKSUM, EXP_IP_CHSUM and EXP_ICMP_CHKSUM are each 4 hex digits
+test_ip_packet() {
+    local inport=$1 hv=$2 eth_src=$3 eth_dst=$4 ipv4_src=$5 ipv4_dst=$6 ip_router=$7 ip_chksum=$8
+    local exp_ip_chksum=$9 exp_icmp_chksum=${10}
+    shift 10
+
+    local ip_ttl=01
+    local packet=${eth_dst}${eth_src}08004500001400004000${ip_ttl}01${ip_chksum}${ipv4_src}${ipv4_dst}
+
+    local reply_icmp_ttl=fe
+    local icmp_type_code_response=0b00
+    local icmp_data=00000000
+    local reply_icmp_payload=${icmp_type_code_response}${exp_icmp_chksum}${icmp_data}
+    local reply=${eth_src}${eth_dst}08004500001c00004000${reply_icmp_ttl}01${exp_ip_chksum}${ip_router}${ipv4_src}${reply_icmp_payload}
+    echo $reply >> vif$inport.expected
+
+    as hv$hv ovs-appctl netdev-dummy/receive vif$inport $packet
+}
+
+ip_to_hex() {
+    printf "%02x%02x%02x%02x" "$@"
+}
+
+for i in 1 2; do
+    net_add n$i
+    ovn-nbctl ls-add sw$i
+
+    sim_add hv$i
+    as hv$i
+    ovs-vsctl add-br br-phys
+    ovn_attach n$i br-phys 192.168.$i.1
+
+    ovn-nbctl lsp-add sw$i sw$i-p${i}0 -- \
+        lsp-set-addresses sw$i-p${i}0 "00:00:00:00:00:0$i 192.168.$i.1"
+
+    ovs-vsctl -- add-port br-int vif$i -- \
+        set interface vif$i \
+        external-ids:iface-id=sw$i-p${i}0 \
+            options:tx_pcap=hv$i/vif$i-tx.pcap \
+            options:rxq_pcap=hv$i/vif$i-rx.pcap \
+            ofport-request=$i
+done
+
+ovn-nbctl lr-add lr0
+for i in 1 2; do
+    ovn-nbctl lrp-add lr0 lrp$i 00:00:00:00:ff:0$i 192.168.$i.254/24
+    ovn-nbctl -- lsp-add sw$i lrp$i-attachment \
+              -- set Logical_Switch_Port lrp$i-attachment type=router \
+                options:router-port=lrp$i addresses='"00:00:00:00:ff:'0$i'"'
+done
+
+OVN_POPULATE_ARP
+# allow some time for ovn-northd and ovn-controller to catch up.
+sleep 1
+
+test_ip_packet 1 1 000000000001 00000000ff01 $(ip_to_hex 192 168 1 1) $(ip_to_hex 192 168 2 1) $(ip_to_hex 192 168 1 254) 0000 7dae f4ff
+OVN_CHECK_PACKETS([hv1/vif1-tx.pcap], [vif1.expected])
+
+OVN_CLEANUP([hv1], [hv2])
+AT_CLEANUP