Connectivity problem (kernel bug?)

From: Dmitry N. Hramtsov (hdn@nsu.ru)
Date: Mon Sep 09 2002 - 02:43:45 EST


Hello All,

I've expiriencing a problem using 802.1q vlan in linux.
It is possible a bug in Linux networking code.
I've posted this problem to linux-vlan list too.

Ok, about the problem in details.
Here is my network scheme:

  __________________________________
 | Smart switch with 802.1Q support |
 |-------------------------------------------------------------+
 | Port1 Port2 Port3 |
 | VLAN1: untagged VLAN1: untagged VLAN1: untagged |
 | VLAN2: 802.1Q VLAN2: 802.1Q VLAN2: no |
 | [*] [*] [*] |
 +--------|---------------------|-------------------|----------+
          | | |
          | | |
        Cisco Linux1 Linux2
  (default gateway)

*** Cisco configuration:

!
interface FastEthernet0/0.1
        ip address 10.0.0.1 255.255.255.0
        encapsulation dot1Q 1 native
!
interface FastEthernet0/0.2
        ip address 10.1.0.1 255.255.255.0
        encapsulation dot1Q 2
!

*** Linux1 configuration

interface eth0
        ip 10.0.0.2
        mask 255.255.255.0
        gateway 10.0.0.1

interface eth0.2
        VLAN ID 2
        ip 10.1.0.2
        mask 255.255.255.0

*** Linux2 configuration

interface eth0
        ip 10.0.0.3
        mask 255.255.255.0
        gateway 10.0.0.1

Now about the problem.
In this scheme all hosts can ping each other by 10.0.0.0/24 ip addresses.
And also cisco and Linux1 can ping each other by 10.1.0.0/24 ip addresses.

But when I am doing "ping 10.1.0.2" from Linux2 i did not receive response.
And even more. Linux1 (10.1.0.2) *sends* no response at all.

I've tried to analyse this and come to a conclusion that this is Linux1 fault.
Well, what is going on:

1. Linux2 (10.0.0.3) send icmp echo request to 10.1.0.2 thru 10.0.0.1
     10.0.0.3 -> 10.0.0.1
2. Cisco receives it and forwards to Linux2 thru VLAN 2
     10.1.0.1 -> 10.1.0.2
3. Linux1 receives icmp request thru eth0.2

All of this stages may be observed using tcpdump on Linux1 and Linux2.

4. Nothing happened any more! ==8-O
No reply sends from Linux1!
Not from eth0 nor from eth0.2!

I am expecting that Linux1 will send reply to 10.0.0.3 using eth0 but it
is not happening.

This bug affects not only icmp packets. No communication can be
established between Linux2 and ip 10.1.0.2 at Linux1.

Any ideas? Comments?

Best regards,
Dmitry N. Hramtsov

p.s. my kernel on Linux1 is 2.4.18

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Sun Sep 15 2002 - 22:00:15 EST