Re: [ovs-dev] [PATCH 3/3] nsh: Add support to compose-packet and use it in system tests.

2024-06-05 Thread Ilya Maximets
On 6/3/24 17:09, Simon Horman wrote:
> On Fri, May 31, 2024 at 11:45:12PM +0200, Ilya Maximets wrote:
>> OVS can parse NSH, but can't compose.  Fix that and get rid of plain
>> hex NSH packets in system tests as they are hard to read or modify.
>>
>> Tcpdump calls modified to write actual pcaps instead of text output,
>> so ovs-pcap can be used while checking the results.
>>
>> While at it, replacing sleeps with more robust waiting for tcpdump
>> to start listening.
> 
> I might have separated a) adding NSH compose support

The tests in the patch provide some test coverage for this as well,
so I didn't want to split for that reason.

> b) more robust tcpdump waiting,

We still need to change '>' to '-w' in very same lines, so I though
it's better to not touch the same code twice and just combine the
change, since it's fairly minor.

> and c) using text rather than hex,
> into 3 patches. But I don't feel strongly about it.

For now, I kept the patches as they are for reasons stated above.
Mostly to avoid touching the same code multiple times.

> 
>> M4 macros are better than shell variables, because we can see the
>> substitution result in the test log.  So, using m4_define and m4_join
>> extensively.
>>
>> Signed-off-by: Ilya Maximets 
> 
> Acked-by: Simon Horman 
> 

Thanks, Simon and Eelco!  I applied the set and backported to all
branches, so we can backport tests for future fixes without trouble
of re-working them for each branch.

Best regards, Ilya Maximets.
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev


Re: [ovs-dev] [PATCH 3/3] nsh: Add support to compose-packet and use it in system tests.

2024-06-05 Thread Eelco Chaudron



On 31 May 2024, at 23:45, Ilya Maximets wrote:

> OVS can parse NSH, but can't compose.  Fix that and get rid of plain
> hex NSH packets in system tests as they are hard to read or modify.
>
> Tcpdump calls modified to write actual pcaps instead of text output,
> so ovs-pcap can be used while checking the results.
>
> While at it, replacing sleeps with more robust waiting for tcpdump
> to start listening.
>
> M4 macros are better than shell variables, because we can see the
> substitution result in the test log.  So, using m4_define and m4_join
> extensively.
>
> Signed-off-by: Ilya Maximets 

Thanks for cleaning this up! This patch looks good to me.

Acked-by: Eelco Chaudron 

___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev


Re: [ovs-dev] [PATCH 3/3] nsh: Add support to compose-packet and use it in system tests.

2024-06-03 Thread Simon Horman
On Fri, May 31, 2024 at 11:45:12PM +0200, Ilya Maximets wrote:
> OVS can parse NSH, but can't compose.  Fix that and get rid of plain
> hex NSH packets in system tests as they are hard to read or modify.
> 
> Tcpdump calls modified to write actual pcaps instead of text output,
> so ovs-pcap can be used while checking the results.
> 
> While at it, replacing sleeps with more robust waiting for tcpdump
> to start listening.

I might have separated a) adding NSH compose support
b) more robust tcpdump waiting, and c) using text rather than hex,
into 3 patches. But I don't feel strongly about it.

> M4 macros are better than shell variables, because we can see the
> substitution result in the test log.  So, using m4_define and m4_join
> extensively.
> 
> Signed-off-by: Ilya Maximets 

Acked-by: Simon Horman 

___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev


[ovs-dev] [PATCH 3/3] nsh: Add support to compose-packet and use it in system tests.

2024-05-31 Thread Ilya Maximets
OVS can parse NSH, but can't compose.  Fix that and get rid of plain
hex NSH packets in system tests as they are hard to read or modify.

Tcpdump calls modified to write actual pcaps instead of text output,
so ovs-pcap can be used while checking the results.

While at it, replacing sleeps with more robust waiting for tcpdump
to start listening.

M4 macros are better than shell variables, because we can see the
substitution result in the test log.  So, using m4_define and m4_join
extensively.

Signed-off-by: Ilya Maximets 
---
 lib/flow.c  |  18 
 tests/system-traffic.at | 177 ++--
 2 files changed, 134 insertions(+), 61 deletions(-)

diff --git a/lib/flow.c b/lib/flow.c
index 8e3402388..dc5fb328d 100644
--- a/lib/flow.c
+++ b/lib/flow.c
@@ -3420,6 +3420,24 @@ flow_compose(struct dp_packet *p, const struct flow 
*flow,
 arp->ar_sha = flow->arp_sha;
 arp->ar_tha = flow->arp_tha;
 }
+} else if (flow->dl_type == htons(ETH_TYPE_NSH)) {
+struct nsh_hdr *nsh;
+
+nsh = dp_packet_put_zeros(p, sizeof *nsh);
+dp_packet_set_l3(p, nsh);
+
+nsh_set_flags_ttl_len(nsh, flow->nsh.flags, flow->nsh.ttl,
+  flow->nsh.mdtype == NSH_M_TYPE1
+  ? NSH_M_TYPE1_LEN : NSH_BASE_HDR_LEN);
+nsh->next_proto = flow->nsh.np;
+nsh->md_type = flow->nsh.mdtype;
+put_16aligned_be32(>path_hdr, flow->nsh.path_hdr);
+
+if (flow->nsh.mdtype == NSH_M_TYPE1) {
+for (size_t i = 0; i < 4; i++) {
+put_16aligned_be32(>md1.context[i], flow->nsh.context[i]);
+}
+}
 }
 
 if (eth_type_mpls(flow->dl_type)) {
diff --git a/tests/system-traffic.at b/tests/system-traffic.at
index c4cebb0a3..3f1a15445 100644
--- a/tests/system-traffic.at
+++ b/tests/system-traffic.at
@@ -8920,21 +8920,29 @@ dnl The flow will encap a nsh header to the TCP syn 
packet
 dnl eth/ip/tcp --> OVS --> eth/nsh/eth/ip/tcp
 AT_CHECK([ovs-ofctl -Oopenflow13 add-flow br0 
"table=0,priority=100,in_port=ovs-p0,ip,actions=encap(nsh(md_type=1)),set_field:0x1234->nsh_spi,set_field:0x11223344->nsh_c1,encap(ethernet),set_field:f2:ff:00:00:00:02->dl_dst,set_field:f2:ff:00:00:00:01->dl_src,ovs-p1"])
 
-NETNS_DAEMONIZE([at_ns1], [tcpdump -l -n -xx -U -i p1 > p1.pcap], 
[tcpdump.pid])
-sleep 1
+NETNS_DAEMONIZE([at_ns1],
+  [tcpdump -l -n -xx -U -i p1 -w p1.pcap 2>tcpdump_err], [tcpdump.pid])
+OVS_WAIT_UNTIL([grep "listening" tcpdump_err])
 
-dnl The hex dump is a TCP syn packet. pkt=eth/ip/tcp
-dnl The packet is sent from p0(at_ns0) interface directed to
-dnl p1(at_ns1) interface
-NS_CHECK_EXEC([at_ns0], [$PYTHON3 $srcdir/sendpkt.py p0 f2 00 00 00 00 02 f2 
00 00 00 00 01 08 00 45 00 00 28 00 01 00 00 40 06 b0 13 c0 a8 00 0a 0a 00 00 
0a 04 00 08 00 00 00 00 c8 00 00 00 00 50 02 20 00 b8 5e 00 00 > /dev/null])
+m4_define([TCP_SYN_PKT], [m4_join([,],
+  [eth_src=f2:00:00:00:00:01,eth_dst=f2:00:00:00:00:02,eth_type=0x0800],
+  [nw_src=192.168.0.10,nw_dst=10.0.0.10],
+  [nw_proto=6,nw_ttl=64,nw_frag=no],
+  [tcp_src=1024,tcp_dst=2048,tcp_flags=syn])])
 
-dnl Check the expected nsh encapsulated packet on the egress interface
-OVS_WAIT_UNTIL([cat p1.pcap | grep -E "0x: *f2ff * *0002 *f2ff * 
*0001 *894f *0fc6" 2>&1 1>/dev/null])
-OVS_WAIT_UNTIL([cat p1.pcap | grep -E "0x0010: *0103 *0012 *34ff *1122 *3344 
* * *" 2>&1 1>/dev/null])
-OVS_WAIT_UNTIL([cat p1.pcap | grep -E "0x0020: * * * *f200 * 
*0002 *f200 *" 2>&1 1>/dev/null])
-OVS_WAIT_UNTIL([cat p1.pcap | grep -E "0x0030: *0001 *0800 *4500 *0028 *0001 
* *4006 *b013" 2>&1 1>/dev/null])
-OVS_WAIT_UNTIL([cat p1.pcap | grep -E "0x0040: *c0a8 *000a *0a00 *000a *0400 
*0800 * *00c8" 2>&1 1>/dev/null])
-OVS_WAIT_UNTIL([cat p1.pcap | grep -E "0x0050: * * *5002 *2000 *b85e 
*" 2>&1 1>/dev/null])
+dnl Send the TCP SYN packet from p0(at_ns0) interface directed to
+dnl p1(at_ns1) interface.
+NS_CHECK_EXEC([at_ns0], [$PYTHON3 $srcdir/sendpkt.py p0 \
+$(ovs-ofctl compose-packet --bare 'TCP_SYN_PKT')], [0], [ignore])
+
+m4_define([NSH_HEADER], [m4_join([,],
+  [eth_src=f2:ff:00:00:00:01,eth_dst=f2:ff:00:00:00:02,eth_type=0x894f],
+  [nsh_ttl=63,nsh_np=3,nsh_spi=0x1234,nsh_si=255],
+  [nsh_mdtype=1,nsh_c1=0x11223344])])
+
+OVS_WAIT_UNTIL([ovs-pcap p1.pcap | grep -q "m4_join([], [^],
+$(ovs-ofctl compose-packet --bare 'NSH_HEADER'),
+$(ovs-ofctl compose-packet --bare 'TCP_SYN_PKT'), [\$])"])
 
 OVS_TRAFFIC_VSWITCHD_STOP
 AT_CLEANUP
@@ -8952,19 +8960,31 @@ dnl The flow will decap a nsh header which in turn 
carries a TCP syn packet
 dnl eth/nsh/eth/ip/tcp --> OVS --> eth/ip/tcp
 AT_CHECK([ovs-ofctl -Oopenflow13 add-flow br0 
"table=0,priority=100,in_port=ovs-p0,dl_type=0x894f, actions=decap(),decap(), 
ovs-p1"])
 
-NETNS_DAEMONIZE([at_ns1], [tcpdump -l -n -xx -U -i p1 > p1.pcap], 
[tcpdump.pid])
-sleep 1
+NETNS_DAEMONIZE([at_ns1],
+