Re: High contention on the sk_buff_head.lock

[Vernon Mauery <vernux@us.ibm.com>]

Eilon Greenstein wrote:
> On Wed, 2009-03-18 at 14:07 -0700, Vernon Mauery wrote:
>>> The real "fix" would be probably to use a multi queue capable NIC
>>> and a NIC driver that sets up multiple queues for TX (normally they
>>> only do for RX). Then cores or a set of cores (often the number
>>> of cores is larger than the number of NIC queues) could avoid this
>>> problem. Disadvantage: more memory use.
>> Hmmm.  So do either the netxen_nic or bnx2x drivers support multiple
>> queues?  (that is the HW that I have access to right now).  And do I
>> need to do anything to set them up?
>>
> The version of bnx2x in net-next support multi Tx queues (and Rx). It
> will open an equal number of Tx and Rx queues up to 16 or the number of
> cores in the system. You can validate that all queues are transmitting
> with "ethtool -S" which has per queue statistics in that version.

Thanks.  I will test to see how this affects this lock contention the
next time the broadcom hardware is available.

--Vernon