[dpdk-dev] [Bug 538] Incorrect packets sent by mlx5 with multiseg configuration in testpmd

bugzilla at dpdk.org bugzilla at dpdk.org
Fri Sep 18 14:24:41 CEST 2020


https://bugs.dpdk.org/show_bug.cgi?id=538

            Bug ID: 538
           Summary: Incorrect packets sent by mlx5 with multiseg
                    configuration in testpmd
           Product: DPDK
           Version: 18.11
          Hardware: All
                OS: All
            Status: UNCONFIRMED
          Severity: major
          Priority: Normal
         Component: testpmd
          Assignee: dev at dpdk.org
          Reporter: david.marchand at redhat.com
  Target Milestone: ---

This is something that Dumitru Ceara caught.

Filing the bz for him as he can't work on a patch right now.

Configuring multisegments in testpmd with the txpkts option, the mlx5 driver
will send truncated packets.

As far as we understand, this is because testpmd does not ask for
DEV_TX_OFFLOAD_MULTI_SEGS offload.
But since the problem is only reproduced with mlx5 driver, this might ask for
more investigation.


# On the tester machine:
$ lspci -vv | grep -i ether | grep -i connectx-5
3b:00.0 Ethernet controller: Mellanox Technologies MT28800 Family [ConnectX-5
Ex]
3b:00.1 Ethernet controller: Mellanox Technologies MT28800 Family [ConnectX-5
Ex]

$ /opt/dpdk-19.11.3/x86_64-native-linuxapp-gcc/app/testpmd -c FF -m 32768 -w
0000:3b:00.0 -w 0000:3b:00.1 -- --txd=256 --rxd=128 -i                          
[...]
testpmd> set txpkts 100,100
testpmd> start tx_first



# On the destination machine:
$ tcpdump -vvv -nn -i ens1f0 -e udp -c 1
tcpdump: listening on ens1f0, link-type EN10MB (Ethernet), capture size 262144
bytes
11:50:42.313110 0c:42:a1:2a:82:49 > 02:00:00:00:00:01, ethertype IPv4 (0x0800),
length 104: truncated-ip - 96 bytes missing! (tos 0x0, ttl 64, id 0, offset 0,
flags [none], proto UDP (17), length 186)
    198.18.0.1.9 > 198.18.0.2.9: [no cksum] UDP, length 158
1 packet captured
32 packets received by filter
0 packets dropped by kernel

-- 
You are receiving this mail because:
You are the assignee for the bug.


More information about the dev mailing list