From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753927Ab1HLLxO (ORCPT ); Fri, 12 Aug 2011 07:53:14 -0400 Received: from mx.ctc-g.co.jp ([131.248.58.1]:33784 "EHLO mx.ctc.ctc-g.co.jp" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751331Ab1HLLxM (ORCPT ); Fri, 12 Aug 2011 07:53:12 -0400 X-Greylist: delayed 1940 seconds by postgrey-1.27 at vger.kernel.org; Fri, 12 Aug 2011 07:53:12 EDT Date: Fri, 12 Aug 2011 20:20:47 +0900 From: "Jun.Kondo" Subject: Request for Redhat To: linux-kernel@vger.kernel.org Cc: "omega-g1@ctc-g.co.jp" , notsuki@redhat.com, "Kozaki, Motokazu" , Hajime Taira Reply-to: jun.kondo@ctc-g.co.jp Message-id: <4E450C8F.4030009@ctc-g.co.jp> MIME-version: 1.0 Content-type: text/plain; charset=ISO-2022-JP Content-transfer-encoding: 7bit X-post-Received: by post01.ctc-g.co.jp (CTC-GN 2006/10/01) id 82FAE74BC; Fri, 12 Aug 2011 20:20:39 +0900 (JST) X-vs: by localhost.is02.ctc-g.co.jp (CTC-GN mail 2009/02/01) id 5592053E61; Fri, 12 Aug 2011 20:20:39 +0900 (JST) X-vs: by is02.ctc-g.co.jp (CTC-GN mail 2009/02/01) id 46D5A53E5E; Fri, 12 Aug 2011 20:20:39 +0900 (JST) User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.2; ja; rv:1.9.2.12) Gecko/20101027 Thunderbird/3.1.6 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org CTC had the following demand; 1. to ensure high throughput from the beginning of tcp connection at normal times by acquiring large default transmission buffer value 2. to limit the block time of the write in order to prevent the timeout of upper layer applications even when the connection has low throughput, such as low rate streaming The root of the issue; 2 can not be achieved with the configuration that satisfies 1. The current behavior is as follows; Write is blocked when tcp transmission buffer (wmem) becomes full. In order to write again after that, one third of the transmission buffer (sk_wmem_queued/2) must be freed. When the throughput is low, timeout occurs by the time when the free buffer space is created, which affects streaming service. The effect of the patch; By putting xxx into the variable yyy, the portion of the transmission buffer becomes zzz, thus timeout will not occur in the low throughput network environment. xxx → integer(e.g. 4) yyy → "sysctl_tcp_lowat" zzz → "sk_wmem_queued >> 4" Also, we think one third of the transmission buffer (sk_wmem_queued/2) is too deterministic, and it should be configurable. -------------------------------------------------- # diff -urN sock.c sock.c.mod --- sock.c 2011-04-14 14:58:03.000000000 +0900 +++ sock.c.mod 2011-04-21 15:31:36.000000000 +0900 @@ -205,6 +205,8 @@ __u32 sysctl_wmem_default = SK_WMEM_MAX; __u32 sysctl_rmem_default = SK_RMEM_MAX; +int sysctl_tcp_lowat = 1; + /* Maximal space eaten by iovec or ancilliary data plus some space */ int sysctl_optmem_max = sizeof(unsigned long)*(2*UIO_MAXIOV + 512); @@ -1005,6 +1007,8 @@ sysctl_wmem_max = 131071; sysctl_rmem_max = 131071; } + + sysctl_tcp_lowat = 1; } /* @@ -2084,4 +2088,5 @@ #ifdef CONFIG_SYSCTL EXPORT_SYMBOL(sysctl_rmem_max); EXPORT_SYMBOL(sysctl_wmem_max); +EXPORT_SYMBOL(sysctl_tcp_lowat); #endif # # diff -urN sock.h sock.h.mod --- sock.h 2011-04-14 14:58:03.000000000 +0900 +++ sock.h.mod 2011-04-21 15:30:07.000000000 +0900 @@ -434,9 +434,12 @@ /* * Compute minimal free write space needed to queue new packets. */ + +extern int sysctl_tcp_lowat; + static inline int sk_stream_min_wspace(struct sock *sk) { - return sk->sk_wmem_queued / 2; + return sk->sk_wmem_queued >> sysctl_tcp_lowat; } static inline int sk_stream_wspace(struct sock *sk) # # diff -urN sysctl_net_core.c sysctl_net_core.c.mod --- sysctl_net_core.c 2011-04-14 14:57:10.000000000 +0900 +++ sysctl_net_core.c.mod 2011-04-04 21:09:48.000000000 +0900 @@ -30,6 +30,7 @@ extern u32 sysctl_xfrm_aevent_rseqth; extern int sysctl_xfrm_larval_drop; extern u32 sysctl_xfrm_acq_expires; +extern int sysctl_tcp_lowat; #endif ctl_table core_table[] = { @@ -151,6 +152,14 @@ .proc_handler = &proc_dointvec }, #endif /* CONFIG_XFRM */ + { + .ctl_name = NET_CORE_TCP_LOWAT, + .procname = "tcp_lowat", + .data = &sysctl_tcp_lowat, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = &proc_dointvec + }, #endif /* CONFIG_NET */ { .ctl_name = NET_CORE_SOMAXCONN, # -------------------------------------------------- ------------------------------------------ Jun.Kondo ITOCHU TECHNO-SOLUTIONS Corporation(CTC) tel:+81-3-6238-6607 fax:+81-3-5226-2369 ------------------------------------------