[dpdk-dev] [PATCH / RFC ] ethdev: Allow rte_eth_dev_configure with zero RX/TX queues

Pattan, Reshma reshma.pattan at intel.com
Fri May 20 10:26:33 CEST 2016



> -----Original Message-----
> From: Simon Kågström [mailto:simon.kagstrom at netinsight.net]
> Sent: Friday, May 20, 2016 7:30 AM
> To: Pattan, Reshma <reshma.pattan at intel.com>; dev at dpdk.org;
> thomas.monjalon at 6wind.com
> Subject: Re: [dpdk-dev] [PATCH / RFC ] ethdev: Allow rte_eth_dev_configure
> with zero RX/TX queues
> 
> Ping? Any more comments on this?
> 

Hi, I don't have any objections, just let's wait if any other comments from committee. 

Thanks,
Reshma

> // Simon
> 
> On 2016-05-16 15:16, Simon Kågström wrote:
> > On 2016-05-16 14:43, Pattan, Reshma wrote:
> >>>> This was added to allow devices,  at least with one direction
> >>>> (RX/TX)
> >>> supported. As, devices with both directions disabled doesn't make  sense
> right?
> >>>
> >>> Well, not for running them, no. But this is a part of the shutdown
> >>> procedure between tests (I should have been more clear I guess).
> >>
> >> Yes I understood this. But I am not sure if you can use
> rte_eth_dev_configure(port, 0, 0) to free the resources.
> >> Can you check if you can use rte_eth_dev_rx_queue_stop/
> >> rte_eth_dev_tx_queue_stop to achieve the same, because they do take care
> of releasing mbufs, but doesn't free the queue's sw-ring and queue.
> >
> > But isn't that very strange behavior. Aren't the descriptor rings
> > allocated in rx_queue_setup()? If so, the sequence
> >
> >   rx_queue_stop(); // Release buffers
> >   rx_queue_start();
> >
> > would leave the descriptor ring empty after start, i.e., not able to
> > receive data.
> >
> > // Simon
> >


More information about the dev mailing list