From mboxrd@z Thu Jan 1 00:00:00 1970 From: annie li Subject: Re: [Xen-devel] [PATCH net] xen-netback: Fix handling of skbs requiring too many slots Date: Wed, 04 Jun 2014 11:59:45 -0400 Message-ID: <538F4271.7080505@oracle.com> References: <1401827406-9341-1-git-send-email-zoltan.kiss@citrix.com> <538F369B.7010100@oracle.com> <538F3E6A.7000002@citrix.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: wei.liu2@citrix.com, ian.campbell@citrix.com, netdev@vger.kernel.org, linux@eikelenboom.it, paul.durrant@citrix.com, david.vrabel@citrix.com, xen-devel@lists.xenproject.org, davem@davemloft.net To: Zoltan Kiss Return-path: Received: from aserp1040.oracle.com ([141.146.126.69]:32490 "EHLO aserp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751010AbaFDP75 (ORCPT ); Wed, 4 Jun 2014 11:59:57 -0400 In-Reply-To: <538F3E6A.7000002@citrix.com> Sender: netdev-owner@vger.kernel.org List-ID: On 2014/6/4 11:42, Zoltan Kiss wrote: > On 04/06/14 16:09, annie li wrote: >> >> On 2014/6/3 16:30, Zoltan Kiss wrote: >>> A recent commit (a02eb4 "xen-netback: worse-case estimate in >>> xenvif_rx_action is >>> underestimating") capped the slot estimation to MAX_SKB_FRAGS, but >>> that triggers >>> the next BUG_ON a few lines down, as the packet consumes more slots >>> than >>> estimated. >>> This patch introduces full_coalesce on the skb callback buffer, which >>> is used in >>> start_new_rx_buffer() to decide whether netback needs coalescing more >>> aggresively. By doing that, no packet should need more than >>> XEN_NETIF_MAX_TX_SIZE / PAGE_SIZE data slots, >> >> (XEN_NETIF_MAX_TX_SIZE+1) / PAGE_SIZE here? > > Do you think about the GSO slot? That's why I wrote "data slot", > however that's probably not a clear terminology. What I mean is: XEN_NETIF_MAX_TX_SIZE is 0xFFFF, and XEN_NETIF_MAX_TX_SIZE / PAGE_SIZE turns out to be 15 slots when PAGE_SIZE is 4096. You was trying to use XEN_NETIF_MAX_TX_SIZE as max size of packet - 64k? > I'll add then that excluding GSO slot, as it doesn't carry data > directly, therefore it's irrelevant from this point of view. Correct.:-) Thanks Annie > > Zoli > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xen.org > http://lists.xen.org/xen-devel