hi, Marvin
see below.
> -----Original Message-----
> From: Liu, Yong
> Sent: January 16, 2017 11:08 AM
> To: Mo, YufengX; dts@dpdk.org
> Cc: Mo, YufengX
> Subject: RE: [dts] [PATCH V1 1/2] Packet capture: upload automation test plan
>
> Yufen,
> Please make sure all lines in test plan and commit are limited to 79 characters.
> And some comments and questions below.
>
> Thanks,
> Marvin
>
> > -----Original Message-----
> > From: dts [mailto:dts-bounces@dpdk.org] On Behalf Of yufengx.mo@intel.com
> > Sent: Monday, January 16, 2017 10:49 AM
> > To: dts@dpdk.org
> > Cc: Mo, YufengX <yufengx.mo@intel.com>
> > Subject: [dts] [PATCH V1 1/2] Packet capture: upload automation test plan
> >
> > From: yufengmx <yufengx.mo@intel.com>
> >
> >
> > Packet capture framework feature support packet capturing on
dpdk ethernet
> > devices.
> > DPDK provides dpdk-pdump tool under app/pdump directory for packet
> > capturing on dpdk.
> >
> > The tool is a Data Plane Development Kit (DPDK) tool that runs as a DPDK
> > secondary process and is capable of enabling packet capture on
dpdk ports.
> > * The
tool depends on libpcap based PMD which is disabled
> > by default in the build configuration files,
> > owing to an external dependency on the
libpcap development files
> > which must be installed on the board.
> > Once the
libpcap development files are installed, the libpcap based PMD
> > can be enabled by setting CONFIG_RTE_LIBRTE_PMD_PCAP=y and
> > recompiling the DPDK.
> >
> > Signed-off-by: yufengmx <yufengx.mo@intel.com>
> > ---
> > test_plans/packet_capture_test_plan.rst | 376
> > ++++++++++++++++++++++++++++++++
> > 1 file changed, 376 insertions(+)
> > create mode 100644
test_plans/packet_capture_test_plan.rst
> >
> > diff --git a/test_plans/packet_capture_test_plan.rst
> > b/test_plans/packet_capture_test_plan.rst
> > new file mode 100644
> > index 0000000..6a51cf9
> > --- /dev/null
> > +++ b/test_plans/packet_capture_test_plan.rst
> > @@ -0,0 +1,376 @@
> > +.. Copyright (c) <2016> Intel Corporation
> > + All rights reserved.
> > +
> > + Redistribution and use in source and binary forms, with or without
> > + modification, are permitted provided that the following conditions
> > + are met:
> > +
> > + - Redistributions of source code must retain the above copyright
> > + notice, this list of conditions and the following disclaimer.
> > +
> > + - Redistributions in binary form must reproduce the above copyright
> > + notice, this list of conditions and the following disclaimer in
> > + the documentation and/or other materials provided with the
> > + distribution.
> > +
> > + - Neither the name of Intel Corporation nor the names of its
> > + contributors may be used to endorse or promote products derived
> > + from this software without specific prior written permission.
> > +
> > + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND
> > CONTRIBUTORS
> > + "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
> > + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
> > + FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
> > + COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT,
> > INDIRECT,
> > + INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
> > + (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
> > OR
> > + SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
> > + HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
> > CONTRACT,
> > + STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
> > + ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
> > + OF THE POSSIBILITY OF SUCH DAMAGE.
> > +
> > +==================
> > + packet capture framework feature
> > +==================
> > +
> > +Packet capture framework feature support packet capturing on
dpdk ethernet
> > devices.
> > +DPDK provides dpdk-pdump tool under app/pdump directory for packet
> > capturing on dpdk.
> > +
> > +The dpdk-pdump application will act as the secondary process. The EAL
> > +thread polls for packet capture
fd. If fd polled matches packet capture
> > +fd, it will initiate packet capture processing.
> > +
> > +The testpmd application will act as the primary process. The primary
> > +process create socket for packet capture connection with the secondary
> > +process and registers socket with packet capture
epoll event. Packet capture
> > event will be polled as part of interrupt thread.
> > +
> > +The primary process creates mempool and two
rte_rings for packets
> > +duplication and sharing packet info with the secondary process respectively.
> > +
> > +Upon receiving packet capture event, the primary process receive either
> > +register RX/TX callbacks or remove RX/TX callbacks message from the
> > +secondary process on socket.If packet matches, reference count of
> > +packet will be incremented and
enqueued to second rte_ring for the secondary
> > process to use.
> > +
> > +Prerequisites
> > +=============
> > +* 2x NICs (2 full duplex ports per NIC) plugged into the available
> > +slots on a platform, another two
nic ports are linked with cables.
> > +
> > +Test cases
> > +=============
> > +The testpmd application act as server process with port-topology
> > +chained mode, the dpdk-pdump act as client process to dump capture
> > +packet with different options setting. Select one port of tester as
tx
> > +port, another port of tester as
rx port, send different type packets
> > +from two ports, check pcap files' content dumped by
scapy and tcpdump to
> > confirm testpmd working correctly, check
pcap files' content dumped by
> > tcpdump and
dpdk-pdump to confirm dpdk-pdump working correctly.
> > +
> > +dpdk-pdump command format
> > +--------------------------
> > +* packet capture framework tool
dpdk-pdump command format, parameters
> > +inside the parenthesis represents the mandatory
parametersm parameters
>
> Typo issue, "parametersm" should be ?
>
> > +inside the square brackets represents optional
> > +parameters::
> > +./x86_64-native-linuxapp-gcc/app/dpdk-pdump -- --pdump= '(port =
> > +<port_id> |device_id = <pci address>), (queue=<queue number>),
> > +(rx-dev=<iface/path to
pcap file> | tx-dev=<iface/path to
pcap file>),
> > +[ring-size=<size>], [mbuf-size=<size>], [total-num-mbufs=<size>]'
> > +
> > +transmission packets' format
> > +--------------------------
> > +* IP_RAW
> > + [Ether()/IP()/Raw('\0'*60)]
> > +* TCP
> > + [Ether()/IP()/TCP()/Raw('\0'*60)]
> > +* UDP
> > + [Ether()/IP()/UDP()/Raw('\0'*60)]
> > +* SCTP
> > + [Ether()/IP()/SCTP()/Raw('\0'*40)]
> > +* IPv6_TCP
> > + [Ether()/IPv6()/TCP()/Raw('\0'*60)]
> > +* IPv6_UDP
> > + [Ether()/IPv6()/UDP()/Raw('\0'*60)]
> > +* IPv6_SCTP
> > + [Ether()/IP()/IPv6()/SCTP()/Raw('\0'*40)]
> > +* VLAN_UDP
> > + [Ether()/Dot1Q()/IP()/UDP()/Raw('\0'*40)]
> > +* TIMESYNC
> > + [Ether(dst='FF:FF:FF:FF:FF:FF',type=0x88f7)/"\\x00\\x02"]
> > +* ARP
> > + [Ether(dst='FF:FF:FF:FF:FF:FF')/ARP()]
> > +* LLDP
> > + [Ether()/LLDP()/LLDPManagementAddress()]
> > + notice: LLDP()/LLDPManagementAddress() method are in dts/dep/lldp.py
> > +
> > +port configuration
> > +--------------------------
> > +* confirm two NICs physical link on a platform:
> > + dut port 0 <---> tester port 0
> > + dut port 1 <---> tester port 1
> > +
> > +* Bind two port on DUT::
> > + ./tools/dpdk_nic_bind.py --bind=igb_uio <dut port 0
pci address> <dut
> > +port 1 pci address>
> > +
> Tools folder has been changed to usertools.
>
> > +* On dut, use port 0 as
rx/tx port.
> > + If
dut port 0 rx dump is set,
scapy send packet from tester port 0
> > +and tcpdump dumps tester
> > + port 1's packet.
> > +
> > + If
dut port 0 tx dump is set,
scapy send packet from tester port 1
> > + and tcpdump dumps tester
port 0's packet.
> > +
> > +* If using interfaces as dpdk-pdump
vdev, prepare two ports on DUT,
> > +which haven't been binded
> > + to
dpdk and have been in linked status
> > +
> > +Test Case 1: test pdump port
> > +=================================
> > +test different port options::
> > +* port=<dut port id>
> > +* device_id=<dut
pci address>
> > +
> > +Boot up dpdk's
testpmd with chained option::
> > + ./x86_64-native-linuxapp-gcc/app/testpmd -c 0x6 -n 4 -- -i --port-
> > topology=chained
> > + testpmd> set
fwd io
> > + testpmd> start
> > +
> Recommend to remove targets which will be change between executions.
>
> > +When test VLAN_UDP type packet transmission, set
vlan:
> > + testpmd>
vlan set filter off 1
> > + testpmd> start
> > +
> > +Boot up dpdk-pdump::
> > + ./x86_64-native-linuxapp-gcc/app/dpdk-pdump -- --pdump
'<port
> > option>,queue=*,\
> > + tx-dev=/tmp/pdump-tx.pcap,rx-dev=/tmp/pdump-rx.pcap'
> > +
> > +Set up linux's
tcpdump to receiver packet on tester::
> > + tcpdump -i <rx port name> -w /tmp/sniff-<rx port
name>.pcap
> > + tcpdump -i <tx port name> -w /tmp/sniff-<tx port
name>.pcap
> > +
> > +Send packet on tester by port 0::
> > + sendp(<packet format>,
iface=<port 0 name>)
> > +
> > +Compare pcap file of
scapy with the pcap file dumped by
tcpdump.
> > +Compare pcap file dumped by
dpdk-pdump with pcap files dumped by
tcpdump.
> > +
> > +Send packet on tester by port 1::
> > + sendp(<packet format>,
iface=<port 1 name>)
> > +
> > +Compare pcap file of
scapy with the pcap file dumped by
tcpdump.
> > +Compare pcap file dumped by
dpdk-pdump with pcap files dumped by
tcpdump.
> > +
> > +Test Case 2: test pdump queue
> > +=================================
> > +test different queue options::
> > +* first queue:
> > + queue=0
> > +* all:
> > + queue=*
>
> Please add some basic description of this parameter. Like "Capture first queue traffic".
> > +
> > +Boot up dpdk's
testpmd with chained option::
> > + ./x86_64-native-linuxapp-gcc/app/testpmd -c 0x6 -n 4 -- -i --port-
> > topology=chained
> > + testpmd> set
fwd io
> > + testpmd> start
> > +
> > +When test VLAN_UDP type packet transmission, set
vlan:
> > + testpmd>
vlan set filter off 1
> > + testpmd> start
> > +
> > +Boot up dpdk-pdump::
> > + ./x86_64-native-linuxapp-gcc/app/dpdk-pdump -- --pdump
'port=0,<queue
> > option>,\
> > + tx-dev=/tmp/pdump-tx.pcap,rx-dev=/tmp/pdump-rx.pcap'
> > +
> > +Set up linux's
tcpdump to receiver packet on tester::
> > + tcpdump -i <rx port name> -w /tmp/sniff-<rx port
name>.pcap
> > + tcpdump -i <tx port name> -w /tmp/sniff-<tx port
name>.pcap
> > +
> > +Send packet on tester by port 0::
> > + sendp(<packet format>,
iface=<port 0 name>)
> > +
> > +Compare pcap file of
scapy with the pcap file dumped by
tcpdump.
> > +Compare pcap file dumped by
dpdk-pdump with pcap files dumped by
tcpdump.
> > +
> > +Send packet on tester by port 1::
> > + sendp(<packet format>,
iface=<port 1 name>)
> > +
> > +Compare pcap file of
scapy with the pcap file dumped by
tcpdump.
> > +Compare pcap file dumped by
dpdk-pdump with pcap files dumped by
tcpdump.
> > +
> > +Test Case 3: test pdump
dev pcap
> > +=================================
>
> Underscore please align with test case name.
>
> > +test different dump options::
> > +* tx-dev=/tmp/pdump-tx.pcap,rx-dev=/tmp/pdump-rx.pcap
> > +* rx-dev=/tmp/pdump-rx.pcap
> > +* tx-dev=/tmp/pdump-tx.pcap
>
> Same as previous comment, add basic descriptions for these parameters.
>
> > +
> > +Boot up dpdk's
testpmd with chained option::
> > + ./x86_64-native-linuxapp-gcc/app/testpmd -c 0x6 -n 4 -- -i --port-
> > topology=chained
> > + testpmd> set
fwd io
> > + testpmd> start
> > +
> > +When test VLAN_UDP type packet transmission, set
vlan:
> > + testpmd>
vlan set filter off 1
> > + testpmd> start
> > +
> > +Boot up dpdk-pdump with
pdump options::
> > + ./x86_64-native-linuxapp-gcc/app/dpdk-pdump -- --pdump
> > 'port=0,queue=*,<dump object>'
> > +
> > +Set up linux's
tcpdump to receiver packet on tester::
> > + tcpdump -i <rx port name> -w /tmp/sniff-<rx port
name>.pcap
> > + tcpdump -i <tx port name> -w /tmp/sniff-<tx port
name>.pcap
> > +
> > +Send packet on tester by port 0::
> > + sendp(<packet format>,
iface=<port 0 name>)
> > +
> > +Compare pcap file of
scapy with the pcap file dumped by
tcpdump.
> > +Compare pcap file dumped by
dpdk-pdump with pcap files dumped by
> > tcpdump(ignore when only set
tx-dev).
> > +
> > +Send packet on tester by port 1::
> > + sendp(<packet format>,
iface=<port 1 name>)
> > +
> > +Compare pcap file of
scapy with the pcap file dumped by
tcpdump.
> > +Compare pcap file dumped by
dpdk-pdump with pcap files dumped by
> > tcpdump(ignore when only set
rx-dev).
> > +
> > +Test Case 4: test pdump
dev iface
> > +=================================
> > +test different dump options::
> > +* tx-dev=<dut
tx port name>,rx-dev=<dut
rx port name>
> > +* rx-dev=<dut
rx port name>
> > +* tx-dev=<dut
tx port name>
> > +
> > +Boot up dpdk's
testpmd with chained option::
> > + ./x86_64-native-linuxapp-gcc/app/testpmd -c 0x6 -n 4 -- -i --port-
> > topology=chained
> > + testpmd> set
fwd io
> > + testpmd> start
> > +
> > +When test VLAN_UDP type packet transmission, set
vlan:
> > + testpmd>
vlan set filter off 1
> > + testpmd> start
> > +
> > +Boot up dpdk-pdump with
pdump options::
> > + ./x86_64-native-linuxapp-gcc/app/dpdk-pdump -- --pdump
> > 'port=0,queue=*,<dump object>'
> > +
> > +Set up linux's
tcpdump to receiver packet on tester::
> > + tcpdump -i <rx port name> -w /tmp/sniff-<rx port
name>.pcap
> > + tcpdump -i <tx port name> -w /tmp/sniff-<tx port
name>.pcap
> > +
> > +Set up linux's
tcpdump to receiver packet of dpdk-pdump on
Dut::
> > + when
rx-dev is set
> > + tcpdump -i <dut
rx port name> -w /tmp/pdump-rx.pcap
> > +
> > + when
tx-dev is set
> > + tcpdump -i <dut
tx port name> -w /tmp/pdump-tx.pcap
> > +
> > +Send packet on tester by port 0::
> > + sendp(<packet format>,
iface=<port 0 name>)
> > +
> > +Compare pcap file of
scapy with the pcap file dumped by
tcpdump.
> > +Compare pcap file dumped by
dpdk-pdump with pcap files dumped by
> > tcpdump(ignore when only set
tx-dev).
> > +
> > +Send packet on tester by port 1::
> > + sendp(<packet format>,
iface=<port 1 name>)
> > +
> > +Compare pcap file of
scapy with the pcap file dumped by
tcpdump.
> > +Compare pcap file dumped by
dpdk-pdump with pcap files dumped by
> > tcpdump(ignore when only set
rx-dev).
> > +
> > +Test Case 5: test pdump ring size
> > +=================================
> > +test ring size option, set value within 2^[1~27]
> > +
> > +Boot up dpdk's
testpmd with chained option::
> > + ./x86_64-native-linuxapp-gcc/app/testpmd -c 0x6 -n 4 -- -i --port-
> > topology=chained
> > + testpmd> set
fwd io
> > + testpmd> start
> > +
> > +When test VLAN_UDP type packet transmission, set
vlan:
> > + testpmd>
vlan set filter off 1
> > + testpmd> start
> > +
> > +Boot up dpdk-pdump with
pdump options::
> > + ./x86_64-native-linuxapp-gcc/app/dpdk-pdump -- --pdump
> > 'port=0,queue=*,\
> > + tx-dev=/tmp/pdump-tx.pcap,rx-dev=/tmp/pdump-rx.pcap,ring-size=1024'
> > +
> > +Set up linux's
tcpdump to receiver packet on tester::
> > + tcpdump -i <rx port name> -w /tmp/sniff-<rx port
name>.pcap
> > + tcpdump -i <tx port name> -w /tmp/sniff-<tx port
name>.pcap
> > +
> > +Send packet on tester by port 0::
> > + sendp(<packet format>,
iface=<port 0 name>)
> > +
> > +Compare pcap file of
scapy with the pcap file dumped by
tcpdump.
> > +Compare pcap file dumped by
dpdk-pdump with pcap files dumped by
tcpdump.
> > +
> > +Send packet on tester by port 1::
> > + sendp(<packet format>,
iface=<port 1 name>)
> > +
> > +Compare pcap file of
scapy with the pcap file dumped by
tcpdump.
> > +Compare pcap file dumped by
dpdk-pdump with pcap files dumped by
tcpdump.
> > +
> > +Test Case 6: test pdump
mbuf size
> > +=================================
> > +test mbuf size option, set value within [252~59520]. min value is
> > +decided by single packet size, max value is decided by test platform
memery
> > size.
>
> Typo issue, should be "memory".
> Should here to verify what will be happened when packet size is over
mbuf size?
[Mo, YufengX]
when size is beyon the range, dpdk-pdump won't boot up. Do you want to add a verification.
will it do the same verification for other range checking?
> Suggest add one case for jumbo frame packet capture, not sure whether jumbo frame should be validated here.
[Mo, YufengX]
Do you mean to write a new case for jumbo frame?
>
> > +
> > +
> > +Boot up dpdk's
testpmd with chained option::
> > + ./x86_64-native-linuxapp-gcc/app/testpmd -c 0x6 -n 4 -- -i --port-
> > topology=chained
> > + testpmd> set
fwd io
> > + testpmd> start
> > +
> > +When test VLAN_UDP type packet transmission, set
vlan:
> > + testpmd>
vlan set filter off 1
> > + testpmd> start
> > +
> > +Boot up dpdk-pdump with
pdump options::
> > + ./x86_64-native-linuxapp-gcc/app/dpdk-pdump -- --pdump
> > 'port=0,queue=*,\
> > + tx-dev=/tmp/pdump-tx.pcap,rx-dev=/tmp/pdump-rx.pcap,mbuf-size=2048'
> > +
> > +Set up linux's
tcpdump to receiver packet on tester::
> > + tcpdump -i <rx port name> -w /tmp/sniff-<rx port
name>.pcap
> > + tcpdump -i <tx port name> -w /tmp/sniff-<tx port
name>.pcap
> > +
> > +Send packet on tester by port 0::
> > + sendp(<packet format>,
iface=<port 0 name>)
> > +
> > +Compare pcap file of
scapy with the pcap file dumped by
tcpdump.
> > +Compare pcap file dumped by
dpdk-pdump with pcap files dumped by
tcpdump.
> > +
> > +Send packet on tester by port 1::
> > + sendp(<packet format>,
iface=<port 1 name>)
> > +
> > +Compare pcap file of
scapy with the pcap file dumped by
tcpdump.
> > +Compare pcap file dumped by
dpdk-pdump with pcap files dumped by
tcpdump.
> > +
> > +Test Case 7: test pdump total
num mbufs
> > +=================================
> > +test total-num-mbufs option, set value within [1025~65535]
>
> What will be affected when change the number?
I think mbuf will be freed after packet saved into
pcap file.
> Will it affect the performance?
[Mo, YufengX]
no things happen when total-num-mbufs changes. Currently, test case will reboot
dpdk-pdump/testpmd
after each different option finish testing.
Within those email discussed with
Reshma, she has mentioned that dpdk-pdump don’t work for performance, you are are in
maillist
too.
>
> > +
> > +Boot up dpdk's
testpmd with chained option::
> > + ./x86_64-native-linuxapp-gcc/app/testpmd -c 0x6 -n 4 -- -i --port-
> > topology=chained
> > + testpmd> set
fwd io
> > + testpmd> start
> > +
> > +When test VLAN_UDP type packet transmission, set
vlan:
> > + testpmd>
vlan set filter off 1
> > + testpmd> start
> > +
> > +Boot up dpdk-pdump with
pdump options::
> > + ./x86_64-native-linuxapp-gcc/app/dpdk-pdump -- --pdump
> > 'port=0,queue=*,\
> > + tx-dev=/tmp/pdump-tx.pcap,rx-dev=/tmp/pdump-rx.pcap,total-num-
> > mbufs=8191'
> > +
> > +Set up linux's
tcpdump to receiver packet on tester::
> > + tcpdump -i <rx port name> -w /tmp/sniff-<rx port
name>.pcap
> > + tcpdump -i <tx port name> -w /tmp/sniff-<tx port
name>.pcap
> > +
> > +Send packet on tester by port 0::
> > + sendp(<packet format>,
iface=<port 0 name>)
> > +
> > +Compare pcap file of
scapy with the pcap file dumped by
tcpdump.
> > +Compare pcap file dumped by
dpdk-pdump with pcap files dumped by
tcpdump.
> > +
> > +Send packet on tester by port 1::
> > + sendp(<packet format>,
iface=<port 1 name>)
> > +
> > +Compare pcap file of
scapy with the pcap file dumped by
tcpdump.
> > +Compare pcap file dumped by
dpdk-pdump with pcap files dumped by
tcpdump.
> > +
> > --
> > 1.9.3