From mboxrd@z Thu Jan 1 00:00:00 1970 From: Blaschka Subject: Re: [RFC] qeth: exploit gro for layer 3 driver Date: Thu, 21 Jan 2010 14:48:45 +0100 Message-ID: <20100121134845.GA28645@tuxmaker.boeblingen.de.ibm.com> References: <20100121123722.GA13815@tuxmaker.boeblingen.de.ibm.com> <20100121.051133.102632014.davem@davemloft.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20100121.051133.102632014.davem@davemloft.net> Sender: netdev-owner@vger.kernel.org List-Archive: List-Post: To: David Miller Cc: netdev@vger.kernel.org, linux-s390@vger.kernel.org List-ID: On Thu, Jan 21, 2010 at 05:11:33AM -0800, David Miller wrote: > From: Blaschka > Date: Thu, 21 Jan 2010 13:37:22 +0100 > > > Since the qeth driver can not use NAPI I would like to hear > > your opinion about following approach to exploit GRO. > > Can you remind us why it can't use NAPI? > > I personally find it hard to believe... :-) Sure, let me try ... on s/390 we neither have have a PCI bus nor a DMA transfer. We have a common software component called qdio which does a DMA like data transfer to/from the hardware. The inbound interface between qeth and qdio is a tasklet running under the control of qdio. So the qeth driver can not disable qdio IRQs (I guess this is a major reason why not using NAPI) and poll the data. qdio is also used for other s/390 non network drivers (e.g. fcp) so changing this component is rather difficult and out of qeth scope. The qdio framework already provides a bottomhalf so todays approach to call netif_rx seems to be questionable anyway ... > -- > To unsubscribe from this list: send the line "unsubscribe netdev" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html