[Bug 231859] ipoib datagram mode: unicast packets are received as multicast packets
bugzilla-noreply at freebsd.org
bugzilla-noreply at freebsd.org
Mon Oct 1 16:34:12 UTC 2018
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=231859
Bug ID: 231859
Summary: ipoib datagram mode: unicast packets are received as
multicast packets
Product: Base System
Version: 11.2-RELEASE
Hardware: amd64
OS: Any
Status: New
Severity: Affects Some People
Priority: ---
Component: kern
Assignee: bugs at FreeBSD.org
Reporter: bmueller at panasas.com
When using the ipoib driver built without IPOIB_CM option, simple commands such
ping do not work. The problem is packets received by the ipoib driver are
being tagged as M_MCAST. Higher layers in the network stack process the packet
as a multicast packet and may ignore it. In the case of 'ping <IB host>', ICMP
will not respond unless the net.inet.icmp.bmcastecho sysctl is set to 0.
Adding debugging to ipoib_input(), the received packet has the following
hwaddr: 0:ff:ff:ff:ff:12:40:1b:ff:ff:0:0:0:0:0:0:ff:ff:ff:ff
The address is a IB multicast address. When I look at the ibdump output on the
sender, the ICMP ECHO packet looks like a unicast packet sent from source LID
15 to destination LID 2.
Extensible Record Format
Timestamp: 0x5bb2469255113055
Record type: 0x15 (Type 21: INFINIBAND)
.001 0101 = Type: INFINIBAND (21)
0... .... = Extension header present: 0
Flags: 0x04 (Capture Interface: 0)
.... ..00 = Capture interface: 0
.... .1.. = Varying record length: 1
.... 0... = Truncated: 0
...0 .... = RX error: 0
..0. .... = DS error: 0
00.. .... = Reserved: 0x0
Record length: 138
Loss counter: 0
Wire length: 122
InfiniBand
Local Route Header
0000 .... = Virtual Lane: 0x0
.... 0000 = Link Version: 0
0000 .... = Service Level: 0
.... 00.. = Reserved (2 bits): 0
.... ..10 = Link Next Header: 0x2
Destination Local ID: 2
0000 0... .... .... = Reserved (5 bits): 0
.... .000 0001 1110 = Packet Length: 30
Source Local ID: 15
Base Transport Header
Opcode: Unreliable Datagram (UD) - SEND only (100)
0... .... = Solicited Event: False
.1.. .... = MigReq: True
..00 .... = Pad Count: 0
.... 0000 = Header Version: 0
Partition Key: 65535
Reserved: 00
Destination Queue Pair: 0x000228
0... .... = Acknowledge Request: False
.000 0000 = Reserved (7 bits): 0
Packet Sequence Number: 203633
DETH - Datagram Extended Transport Header
Queue Key: 0x0000000000000b1b
Reserved: 00
Source Queue Pair: 0x00000210
IBA Payload - appears to be EtherType encapsulated
Ethertype: IPv4 (0x0800)
Reserved: 0000
Invariant CRC: 0x29a166b6
Variant CRC: 0x39c8
Internet Protocol Version 4, Src: 192.168.1.204, Dst: 192.168.1.116
0100 .... = Version: 4
.... 0101 = Header Length: 20 bytes (5)
Differentiated Services Field: 0x00 (DSCP: CS0, ECN: Not-ECT)
0000 00.. = Differentiated Services Codepoint: Default (0)
.... ..00 = Explicit Congestion Notification: Not ECN-Capable Transport
(0)
Total Length: 84
Identification: 0xd946 (55622)
Flags: 0x4000, Don't fragment
0... .... .... .... = Reserved bit: Not set
.1.. .... .... .... = Don't fragment: Set
..0. .... .... .... = More fragments: Not set
...0 0000 0000 0000 = Fragment offset: 0
Time to live: 64
Protocol: ICMP (1)
Header checksum: 0xdcd1 [validation disabled]
[Header checksum status: Unverified]
Source: 192.168.1.204
Destination: 192.168.1.116
Internet Control Message Protocol
Type: 8 (Echo (ping) request)
Code: 0
Checksum: 0x1ea9 [correct]
[Checksum Status: Good]
Identifier (BE): 10727 (0x29e7)
Identifier (LE): 59177 (0xe729)
Sequence number (BE): 1 (0x0001)
Sequence number (LE): 256 (0x0100)
Timestamp from icmp data: Oct 1, 2018 16:08:50.000000000 GMT
[Timestamp from icmp data (relative): 0.332293530 seconds]
Data (48 bytes)
0000 a7 f9 04 00 00 00 00 00 10 11 12 13 14 15 16 17 ................
0010 18 19 1a 1b 1c 1d 1e 1f 20 21 22 23 24 25 26 27 ........ !"#$%&'
0020 28 29 2a 2b 2c 2d 2e 2f 30 31 32 33 34 35 36 37 ()*+,-./01234567
Data: a7f9040000000000101112131415161718191a1b1c1d1e1f...
[Length: 48]
--
You are receiving this mail because:
You are the assignee for the bug.
More information about the freebsd-bugs
mailing list