From mboxrd@z Thu Jan 1 00:00:00 1970 From: Pankaj Gupta Subject: Re: [PATCH net-next 1/4] net: allow large number of rx queues Date: Thu, 20 Nov 2014 11:31:28 -0500 (EST) Message-ID: <497267937.1868067.1416501088558.JavaMail.zimbra@redhat.com> References: <1416327778-17716-1-git-send-email-pagupta@redhat.com> <1416327778-17716-2-git-send-email-pagupta@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Cc: linux-kernel@vger.kernel.org, netdev , David Miller , Jason Wang , "Michael S. Tsirkin" , dgibson@redhat.com, vfalico@gmail.com, Eric Dumazet , Vladislav Yasevich , Jerry Chu , wuzhy@redhat.com, Pavel Emelyanov , Tom Herbert , bhutchings@solarflare.com, xii@google.com, Stephen Hemminger , =?utf-8?B?SmnFmcOtIFDDrXJrbw==?= , sergei shtylyov To: Cong Wang Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org > On Tue, Nov 18, 2014 at 8:22 AM, Pankaj Gupta wrote: > > > > As vmalloc() adds overhead on a critical network path, > > __GFP_REPEAT flag is used with kzalloc() to do this fallback > > only when really needed. > > > > Are you sure we need __GFP_REPEAT? We have vmalloc() as > fallback of kmalloc() in many places of networking (grep kvfree), > none of those I checked has this flag set. Its there in netif_alloc_netdev_queues() function in same file. Also, I found it some other places as well. I think its good to have. >