Hi Stefano

On Sep 29, 2012, at 6:33 PM, Stefano Debenedetti <[email protected]> wrote:

> Hi all,
> 
> I am using PFRING DNA DAQ module for snort on a 32-core machine with
> a dual-port 82599 card and I noticed that even if 32 queues are
> enabled [1], only 16 of them actually receive interrupts and packets.
> 
>> From the ixgbe DNA driver source code (IXGBE_MAX_RSS_INDICES) and
> from a metaflows post [2] it seems that this is a hardware
> limitation. On the other hand the ixgbe driver README file says that
> 82599 supports up to 64 queues but is probably referring to
> IXGBE_MAX_MSIX_Q_VECTORS.
> 
> Is it a hardware limitation?

Yes, RSS supports up to 16 queues on 82598/82599.

> 
> If it is a hardware limitation, why 32 queues are created if only 16
> can be used? I.e. what's the use of the MSIX_Q_VECTORS that exceed
> the RSS_INDICES number?

The other queues (up to 64 on 82598, 128 on 82599) can still be used by the 
Flow Director (hw filters), please have a look at:
PF_RING/userland/examples/pfcount_82599.c

Ciao
Alfredo

> 
> thanks ciao
> ste
> 
> [1]
> 
> ixgbe 0000:21:00.0: dna0: Enabled Features: RxQ: 32 TxQ: 32 FdirHash LRO
> ixgbe 0000:21:00.1: dna1: Enabled Features: RxQ: 32 TxQ: 32 FdirHash LRO
> 
> 
> [2]
> 
> https://www.metaflows.com/technology/10-gbps-pf_ring-2/
> 
> _______________________________________________
> Ntop-misc mailing list
> [email protected]
> http://listgateway.unipi.it/mailman/listinfo/ntop-misc

_______________________________________________
Ntop-misc mailing list
[email protected]
http://listgateway.unipi.it/mailman/listinfo/ntop-misc

Reply via email to