Bug 196512

Summary: VLANs, tg3 driver, and 2.6.9-34.EL kernel update
Product: Red Hat Enterprise Linux 4 Reporter: Robin Yamaguchi <rhy>
Component: kernelAssignee: Jason Baron <jbaron>
Status: CLOSED DUPLICATE QA Contact: Brian Brock <bbrock>
Severity: medium Docs Contact:
Priority: medium    
Version: 4.0CC: knoel
Target Milestone: ---   
Target Release: ---   
Hardware: i686   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2006-06-28 03:21:47 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On:    
Bug Blocks: 181409    

Description Robin Yamaguchi 2006-06-23 20:30:02 UTC
Description of problem:

Since upgrading our RHEL4 boxes to the 2.6.9-34.EL kernel, we've experienced
problems with our virtual interfaces and VLAN tagged frames.  Before upgrading
to the kernel, our multi-homed box worked fine, dhcping for addresses on each
virtual interface.

Our two interfaces are eth0.75 and eth0.76, using a tg3 driver for an onboard
Broadcom Corporation NetXtreme BCM5751 Gigabit NIC.

After the kernel upgrade to 2.6.9-34.EL, eth0.75 is unable to DHCP for an IP
address.  The DHCP server never sees the DHCPDISCOVER.

However, eth0.76, also using tagged frames, is able to successfully DHCP for an
address.

Strangely, when I disable eth0.75 (using ONBOOT=no in ifcfg-eth0.75), and
restart the network service, eth0.76 can no longer DHCP.  Again, the DHCP server
never sees the DHCPDISCOVER.

I tried statically assigned an IP address, but the problemed interface still
wasn't able to see the network.  However, I am able to tcpdump, and I see the
apropriate ARP requests acrossed that tagged network.

I believe there is something wrong with the tg3 driver we are using for our NIC.
 I installed a PCI 3com NIC, setup virtual interfaces simularly, connected to
the same switch port as I was previously using, and virtual interfaces DHCP fine
with the latest kernels.

However, I tried downloading the source code for the latest tg3 driver from
broadcomm's website.  It compiled fine, but I continued to have the same problem. 

Version-Release number of selected component (if applicable):
2.6.9-34.EL and 2.6.9-34.0.1.EL.  This setup was working using 2.6.9-22.0.2.EL
and previous kernels.
 
How reproducible: Always

Here are the relevant files:

# more /etc/sysconfig/network
NETWORKING=yes
HOSTNAME=localhost.localdomain
NISDOMAIN=<our_nis_domain>
GATEWAY=<our_gateway_ip>
VLAN=yes

# more /etc/sysconfig/network-scripts/ifcfg-eth0
# Broadcom Corporation NetXtreme BCM5751 Gigabit Ethernet PCI Express
DEVICE=eth0
ONBOOT=no
BOOTPROTO=none
PEERDNS=no
TYPE=Ethernet

# more /etc/sysconfig/network-scripts/ifcfg-eth0.75 
DEVICE=eth0.75
BOOTPROTO=dhcp
ONBOOT=yes
PEERDNS=no

# more /etc/sysconfig/network-scripts/ifcfg-eth0.76 
DEVICE=eth0.76
BOOTPROTO=dhcp
ONBOOT=yes
PEERDNS=no

Comment 1 Jason Baron 2006-06-26 14:18:07 UTC
hmmm, can you try the rhel4 U4 beta kernel, thanks.

http://people.redhat.com/~jbaron/rhel4/

Comment 2 Robin Yamaguchi 2006-06-26 16:56:12 UTC
Installing 2.6.9-34.4.EL beta kernel seems to have fixed my problem!  Thank you!

I'm curious to know, what was the issue?

Comment 3 Jason Baron 2006-06-28 03:18:42 UTC
34.4? shouldn't it be -40? just curious what kernel you tried. I belive the fix
is bugzilla 188087, but i can double check for you if its really important.
thatnks for testing this for us :)

*** This bug has been marked as a duplicate of 188087 ***

Comment 4 Jason Baron 2006-06-28 03:21:47 UTC

*** This bug has been marked as a duplicate of 188087 ***

Comment 5 Robin Yamaguchi 2006-06-28 04:46:41 UTC
[root@crosby Desktop]# uname -a
Linux crosby.pcs 2.6.9-39.4.EL #1 Fri Jun 23 17:21:50 EDT 2006 i686 i686 i386 
GNU/Linux
[root@crosby Desktop]# more /etc/redhat-release
Red Hat Enterprise Linux ES release 4 (Nahant Update 3)