From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jim Baxter Subject: Re: [PATCH v1 3/3] usb: gadget: NCM: Stop RX TCP Bursts getting dropped. Date: Fri, 30 May 2014 12:25:48 +0100 Message-ID: <53886ABC.8030303@mentor.com> References: <1401383543-10181-1-git-send-email-jim_baxter@mentor.com> <1401383543-10181-4-git-send-email-jim_baxter@mentor.com> <1401390263.3645.51.camel@edumazet-glaptop2.roam.corp.google.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Cc: Felipe Balbi , Greg Kroah-Hartman , , , =?UTF-8?B?QmrDuA==?= =?UTF-8?B?cm4gTW9yaw==?= , David Laight , To: Eric Dumazet Return-path: In-Reply-To: <1401390263.3645.51.camel@edumazet-glaptop2.roam.corp.google.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org On 29/05/14 20:04, Eric Dumazet wrote: > On Thu, 2014-05-29 at 18:12 +0100, Jim Baxter wrote: >> This fixes a problem with dropped packets over 16k CDC-NCM >> when the connection is being heavily used. >> >> The issue was that the skb truesize for the unpacked NCM >> packets was too high after they were cloned from the 16k >> skb, this lead to the potential memory calculated by the >> Kernel running out of memory earlier then it should. >> >> Signed-off-by: Jim Baxter >> --- > > Note the patch is OK, but changelog a bit misleading ;) > > Kernel was not running out of memory, because truesize was correct. > > The problem here is that a frame was consuming more kernel memory than > really needed, so chances of hitting socket sk_rcvbuf limit was high. > > BTW : > #define NTB_OUT_SIZE 16384 > > alloc_skb(size) -> > kmalloc(16384 + sizeof(struct skb_shared_info)) -> > roundup() => 32768 > > So truesize of the skb was infact ~32KB, which is really insane indeed. > After your patch, its back to ~2KB > > Acked-by: Eric Dumazet > > > Thank you, I will clarify the description. Jim