Has anyone had success running VLANs nicely on a Fedora Core 2 box? I had a number of kernel errors dumping register info on me. Background: The box is a Celeron 700 with 2 Intel Dual Port Nics (4 ethernet interfaces total). I'm using the e100 driver. The kernel is a stock Fedora 2.6.1-521 kernel, all recent updates. I'm using vconfig 1.8. This box is also running squid with a smallish cache (10 gigs). It's been rock solid stable routing without vlans. I'm looking to expand the last interface out to use vlans as we're adding more segments and I want to do it without more NICs. I'm also running snmpd and doing some basic packet shaping with tc (1 queue on one interface, choking off a small segment's speed). Here's what happened yesterday as I was trying things on this system: modprobe 8021q worked fine, as did the vconfig. I added a single vlan to my eth3 device. I used ifconfig to bring up the vlan, and that worked fine. I was able to ping okay (connecting to a Cisco switch to a trunked port and pinged through to device connected on another port with vlan membership). I ran ifconfig again with no arguments to see my status of all my interfaces. It was stuck--it wouldn't respond my keyboard input, ctrl-c did nothing. I thought I had hung the box, but I could open up a second ssh shell to it without difficulty. Looks like it was an oops, here's my system log of the pertinent info: Nov 14 12:15:12 routerhostname kernel: Unable to handle kernel NULL pointer dereference at virtual address 00000000 Nov 14 12:15:12 routerhostname kernel: printing eip: Nov 14 12:15:12 routerhostname kernel: 00000000 Nov 14 12:15:12 routerhostname kernel: *pde = 00000000 Nov 14 12:15:12 routerhostname kernel: Oops: 0000 [#1] Nov 14 12:15:12 routerhostname kernel: Modules linked in: sch_tbf 8021q md5 ipv6 sd_mod scsi_mod e100 mii dm_mod uhci_hcd button battery asus_acpi ac ext3 jbd Nov 14 12:15:12 routerhostname kernel: CPU: 0 Nov 14 12:15:12 routerhostname kernel: EIP: 0060:[<00000000>] Not tainted Nov 14 12:15:12 routerhostname kernel: EFLAGS: 00010246 (2.6.8-1.521) Nov 14 1routerhostname2:15:12 routerhostname kernel: EIP is at 0x0 Nov 14 12:15:12 routerhostname kernel: eax: 00000000 ebx: 10a3f4e8 ecx: 00000000 edx: 00000000 Nov 14 12:15:12 routerhostname kernel: esi: 0f3c8ef0 edi: 0f3c8ef0 ebp: 00000000 esp: 0f3c8eac Nov 14 12:15:12 routerhostname kernel: ds: 007b es: 007b ss: 0068 Nov 14 12:15:12 routerhostname kernel: Process snmpd (pid: 1290, threadinfo=0f3c8000 task=0e9738f0) Nov 14 12:15:12 routerhostname kernel: Stack: 1281c4c1 00000000 00000000 0f3c8ee0 128421ea 0f3c8ef0 0f3c8f30 128421f9 Nov 14 12:15:12 routerhostname kernel: 00000000 128571f8 ffffffa1 00008947 10a3f000 33687465 00000000 00000000 Nov 14 12:15:12 routerhostname kernel: 00000000 00000000 f0ab9ab3 00000000 00000000 10a3f000 00008947 ffffffed Nov 14 12:15:12 routerhostname kernel: Call Trace: Nov 14 12:15:12 routerhostname kernel: [<1281c4c1>] generic_mii_ioctl+0x79/0x140 [mii] Nov 14 12:15:12 routerhostname kernel: [<128421ea>] e100_do_ioctl+0x0/0x11 [e100] Nov 14 12:15:12 routerhostname kernel: [<128421f9>] e100_do_ioctl+0xf/0x11 [e100] Nov 14 12:15:12 routerhostname kernel: [<128571f8>] vlan_dev_ioctl+0x85/0xbe [8021q] Nov 14 12:15:12 routerhostname kernel: [<0229f92f>] dev_ifsioc+0x304/0x310 Nov 14 12:15:12 routerhostname kernel: [<12857173>] vlan_dev_ioctl+0x0/0xbe [8021q] Nov 14 12:15:12 routerhostname kernel: [<0229faa7>] dev_ioctl+0x16c/0x28a Nov 14 12:15:12 routerhostname kernel: [<022d377b>] udp_ioctl+0x0/0x190 Nov 14 12:15:12 routerhostname kernel: [<022d967b>] inet_ioctl+0x6e/0x73 Nov 14 12:15:12 routerhostname kernel: [<02296a84>] sock_ioctl+0x2f4/0x3aa Nov 14 12:15:12 routerhostname kernel: [<021752d2>] sys_ioctl+0x29a/0x33c Nov 14 12:15:12 routerhostname kernel: [<02160c5e>] sys_read+0x3c/0x62 Nov 14 12:15:12 routerhostname kernel: Code: Bad EIP value. Nov 14 12:15:16 routerhostname kernel: <7>eth3.20: no IPv6 routers present I thought it >might< have something to do with the IPv6 stuff loading, and I really don't need that (though I would like to be able to use it someday), though above seems to indicate snmpd is where the issue is. I figured I would disable IPv6. A little google action and I determined that could be disabled by adding the following: alias net-pf-10 off to /etc/modprobe.conf I rebooted, and was able to run ifconfig twice without issue. I thought I had it licked. Then I applied my tc queuing discipline to another interface (not using VLANs, just a native interface). I frequently issue the following command to see how it's ticking along in semi-real time: watch tc -s -d qdisc ls And this too, would hang just like the ifconfig command. Is this just a coincidence that it happens during an SNMP poll? At what appeared to be the same time, the kernel did an oops like this, nearly same as above (note lack of IPv6 stuff loaded): Nov 14 12:55:04 routerhostname snmpd[1272]: Received SNMP packet(s) from xxx.xxx.xxx.xxx Nov 14 12:55:04 routerhostname kernel: Unable to handle kernel NULL pointer dereference at virtual address 00000000 Nov 14 12:55:04 routerhostname kernel: printing eip: Nov 14 12:55:04 routerhostname kernel: 00000000 Nov 14 12:55:04 routerhostname kernel: *pde = 00000000 Nov 14 12:55:04 routerhostname kernel: Oops: 0000 [#1] Nov 14 12:55:04 routerhostname kernel: Modules linked in: sch_tbf 8021q sd_mod scsi_mod e100 mii dm_mod uhci_hcd button battery asus_acpi ac ext3 jbd Nov 14 12:55:04 routerhostname kernel: CPU: 0 Nov 14 12:55:04 routerhostname kernel: EIP: 0060:[<00000000>] Not tainted Nov 14 12:55:04 routerhostname kernel: EFLAGS: 00010246 (2.6.8-1.521) Nov 14 12:55:04 routerhostname kernel: EIP is at 0x0 Nov 14 12:55:04 routerhostname kernel: eax: 00000000 ebx: 10dab8e8 ecx: 00000000 edx: 00000000 Nov 14 12:55:04 routerhostname kernel: esi: 0f13aef0 edi: 0f13aef0 ebp: 00000000 esp: 0f13aeac Nov 14 12:55:04 routerhostname kernel: ds: 007b es: 007b ss: 0068 Nov 14 12:55:04 routerhostname kernel: Process snmpd (pid: 1272, threadinfo=0f13a000 task=0f1d9910) Nov 14 12:55:04 routerhostname kernel: Stack: 1281c4c1 00000000 00000000 0f13aee0 128421ea 0f13aef0 0f13af30 128421f9 Nov 14 12:55:04 routerhostname kernel: 00000000 1287b1f8 ffffffa1 00008947 10dab400 33687465 00000000 00000000 Nov 14 12:55:04 routerhostname kernel: 00000000 00000000 f0ab9ab3 00000000 00000000 10dab400 00008947 ffffffed Nov 14 12:55:04 routerhostname kernel: Call Trace: Nov 14 12:55:04 routerhostname kernel: [<1281c4c1>] generic_mii_ioctl+0x79/0x140[mii] Nov 14 12:55:04 routerhostname kernel: [<128421ea>] e100_do_ioctl+0x0/0x11 [e100] Nov 14 12:55:04 routerhostname kernel: [<128421f9>] e100_do_ioctl+0xf/0x11 [e100] Nov 14 12:55:04 routerhostname kernel: [<1287b1f8>] vlan_dev_ioctl+0x85/0xbe [8021q] Nov 14 12:55:04 routerhostname kernel: [<0229f92f>] dev_ifsioc+0x304/0x310 Nov 14 12:55:04 routerhostname kernel: [<1287b173>] vlan_dev_ioctl+0x0/0xbe [8021q] Nov 14 12:55:04 routerhostname kernel: [<0229faa7>] dev_ioctl+0x16c/0x28a Nov 14 12:55:04 routerhostname kernel: [<022d377b>] udp_ioctl+0x0/0x190 Nov 14 12:55:04 routerhostname kernel: [<022d967b>] inet_ioctl+0x6e/0x73 Nov 14 12:55:04 routerhostname kernel: [<02296a84>] sock_ioctl+0x2f4/0x3aa Nov 14 12:55:04 routerhostname kernel: [<021752d2>] sys_ioctl+0x29a/0x33c Nov 14 12:55:04 routerhostname kernel: [<02160c5e>] sys_read+0x3c/0x62 Nov 14 12:55:04 routerhostname kernel: Code: Bad EIP value. Thoughts? Steve
