From mboxrd@z Thu Jan 1 00:00:00 1970 From: Yuval Shaia Subject: Re: [PATCH v3] IB/ipoib: Scatter-Gather support in connected mode Date: Tue, 12 May 2015 09:08:16 +0300 Message-ID: <20150512060815.GA2668@yuval-lab> References: <1431338642-22812-1-git-send-email-yuval.shaia@oracle.com> <1431344407.25060.29.camel@opteya.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Content-Disposition: inline In-Reply-To: <1431344407.25060.29.camel-RlY5vtjFyJ3QT0dZR+AlfA@public.gmane.org> Sender: linux-rdma-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Yann Droneaud Cc: john.sobecki-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org, christian-x8YiTEMPTCFhl2p70BpVqQ@public.gmane.org, dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org, linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-Id: linux-rdma@vger.kernel.org On Mon, May 11, 2015 at 01:40:07PM +0200, Yann Droneaud wrote: > Hi, >=20 > Le lundi 11 mai 2015 =E0 03:04 -0700, Yuval Shaia a =E9crit : > > By default, IPoIB-CM driver uses 64k MTU. Larger MTU gives better p= erformance. > > This MTU plus overhead puts the memory allocation for IP based pack= ets at 32 > > 4k pages (order 5), which have to be contiguous. > > When the system memory under pressure, it was observed that allocat= ing 128k > > contiguous physical memory is difficult and causes serious errors (= such as > > system becomes unusable). > >=20 > > This enhancement resolve the issue by removing the physically conti= guous memory > > requirement using Scatter/Gather feature that exists in Linux stack= =2E > >=20 > > With this fix Scatter-Gather will be supported also in connected mo= de. > >=20 > > This change reverts the some of the change made in commit > > e112373fd6aa280bd2cbc0d5cc3809115325a1be > > ("IPoIB/cm: Reduce connected mode TX object size)". > >=20 > > The ability to use SG in IPoIB CM is possible because the coupling > > between NETIF_F_SG and NETIF_F_CSUM was removed in commit > > ec5f061564238892005257c83565a0b58ec79295 > > ("net: Kill link between CSUM and SG features.") > >=20 > > Signed-off-by: Yuval Shaia > > --- > > v2: Enhance commit log message > >=20 > > I'd like to thank John Sobecki and > > Christian Marie for helping me with reviewing= this patch. > > I will be glad if you can ACK it. > >=20 > > v3: Add an empty line to start a new paragraph and use 12 hex-digit= s identifier >=20 > Thanks, but you don't have to use the full commit identifier, 12 > characters were enough. Will do. >=20 > Anyway, you haven't addressed the comments embedded in the patch in m= y > reply [1] to your v2 patch. Changing comments was quite easy, the rest will take a bit more, v4 wil= l address it. >=20 > [1] > http://marc.info/?i=3D1431335754.25060.26.camel-RlY5vtjFyJ3QT0dZR+AlfA@public.gmane.org > http://mid.gmane.org/1431335754.25060.26.camel-RlY5vtjFyJ3QT0dZR+AlfA@public.gmane.org >=20 > Regards. >=20 > --=20 > Yann Droneaud > OPTEYA >=20 >=20 -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" i= n the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html