From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932445Ab1LOJf0 (ORCPT ); Thu, 15 Dec 2011 04:35:26 -0500 Received: from mailhub.sw.ru ([195.214.232.25]:18196 "EHLO relay.sw.ru" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758373Ab1LOJfY (ORCPT ); Thu, 15 Dec 2011 04:35:24 -0500 From: Glauber Costa To: davem@davemloft.net Cc: linux-kernel@vger.kernel.org, netdev@vger.kernel.org, cgroups@vger.kernel.org Subject: [PATCH 0/2] Proposed fixes for tcp memory pressure Date: Thu, 15 Dec 2011 13:34:30 +0400 Message-Id: <1323941672-14324-1-git-send-email-glommer@parallels.com> X-Mailer: git-send-email 1.7.6.4 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi, I propose the following two fixes for the issues we're having in linux-next. The first one just moves the definition out of the CONFIG_INET, and should fix the compile problems with allnoconfig. The second one is a bit bigger, and takes away the socket registration functions. As I've explained in the Changelog, I believe this to be the saner way to do it. We need to do allocations in both create and init time. This was compile tested with 4 different randconfigs, CGROUPS off, KMEM cgroup off, INET off, all them on, and boot tested with everything on. Thanks. Glauber Costa (2): Move limit definitions outside CONFIG_INET Explicitly call tcp creation and init from memcontrol.c include/linux/memcontrol.h | 2 +- include/net/sock.h | 2 -- include/net/tcp_memcontrol.h | 19 ++++++++++++++++++- mm/memcontrol.c | 13 ++++--------- net/core/sock.c | 37 ------------------------------------- net/ipv4/tcp_memcontrol.c | 11 ++++++----- 6 files changed, 29 insertions(+), 55 deletions(-) -- 1.7.6.4