On 11/15/2013 04:09 PM, Andrey Korolyov wrote:
On 11/15/2013 06:58 PM, Thomas Graf wrote:
> On 11/14/2013 07:01 PM, Andrey Korolyov wrote:
>> Because of my ugly english Matt understood me a bit incorrect - vlan
>> splinters are okay but table above from
redhat.com suggests that almost
>> everything should work even without specifying a set of VLAN tags and
>> vlan-splinters at all which is seemingly not true for our lab tests with
>> same RDO kernel. So question is if we can avoid using any hacks for
>> specified drivers in the upcoming releases of RDO kernel by small amount
>> of patchwork or not.
>
> The only way to fix this is to fix individual drivers or to enable
> splinters to trigger vlan group membership.
>
> Could you specify what you mean by saying "specifying a set of VLAN
> tags"? Are you referring to the requirement to tag at least one port
> for each VID you are using in order for splinters to be effective?
>
Yes, exactly. We have a lot of reports from users that the some bcm
cards w/ bnx and bnx2x, especially UCS chassis does not work with vlans
without splinters. Unfortunately there is no such hardware around to
reproduce and test and at least emulated e1000 requires splinters to
work. Will follow-up conversation with results from ten gigabit bcms in
a couple of days.
e1000 is a known problem. It is being worked on.
Please let us know as soon as you have confirmation for bnx2x. It's
known good driver based on our testing which we enabled specifically.
Filing a BZ for each individual failing driver will expedite the
process on our side.