From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752866AbaHTT02 (ORCPT ); Wed, 20 Aug 2014 15:26:28 -0400 Received: from mx1.redhat.com ([209.132.183.28]:13411 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751425AbaHTT01 (ORCPT ); Wed, 20 Aug 2014 15:26:27 -0400 Date: Wed, 20 Aug 2014 21:23:30 +0200 From: Oleg Nesterov To: Andrew Morton , KAMEZAWA Hiroyuki , David Rientjes , KOSAKI Motohiro Cc: Alexander Viro , Cyrill Gorcunov , "Eric W. Biederman" , "Kirill A. Shutemov" , Peter Zijlstra , Hugh Dickins , linux-kernel@vger.kernel.org Subject: [PATCH 3/4] mempolicy: sanitize the usage of get_task_policy() Message-ID: <20140820192330.GD8524@redhat.com> References: <20140805194627.GA30693@redhat.com> <20140820192207.GA8524@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20140820192207.GA8524@redhat.com> User-Agent: Mutt/1.5.18 (2008-05-17) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Cleanup + preparation. Every user of get_task_policy() calls it unconditionally, even if it is not going to use the result. get_task_policy() is cheap but still this does not look clean, plus the code looks simpler if get_task_policy() is called only when this is really needed. Note: I hope this is correct, but it is not clear why vma_policy_mof() doesn't fall back to get_task_policy() if ->get_policy() returns NULL. Signed-off-by: Oleg Nesterov --- mm/mempolicy.c | 25 ++++++++++++++----------- 1 files changed, 14 insertions(+), 11 deletions(-) diff --git a/mm/mempolicy.c b/mm/mempolicy.c index 656db97..b86b08e 100644 --- a/mm/mempolicy.c +++ b/mm/mempolicy.c @@ -1621,14 +1621,11 @@ COMPAT_SYSCALL_DEFINE6(mbind, compat_ulong_t, start, compat_ulong_t, len, struct mempolicy *get_vma_policy(struct task_struct *task, struct vm_area_struct *vma, unsigned long addr) { - struct mempolicy *pol = get_task_policy(task); + struct mempolicy *pol = NULL; if (vma) { if (vma->vm_ops && vma->vm_ops->get_policy) { - struct mempolicy *vpol = vma->vm_ops->get_policy(vma, - addr); - if (vpol) - pol = vpol; + pol = vma->vm_ops->get_policy(vma, addr); } else if (vma->vm_policy) { pol = vma->vm_policy; @@ -1643,12 +1640,15 @@ struct mempolicy *get_vma_policy(struct task_struct *task, } } + if (!pol) + pol = get_task_policy(task); + return pol; } bool vma_policy_mof(struct task_struct *task, struct vm_area_struct *vma) { - struct mempolicy *pol = get_task_policy(task); + struct mempolicy *pol = NULL; if (vma) { if (vma->vm_ops && vma->vm_ops->get_policy) { @@ -1660,11 +1660,14 @@ bool vma_policy_mof(struct task_struct *task, struct vm_area_struct *vma) mpol_cond_put(pol); return ret; - } else if (vma->vm_policy) { - pol = vma->vm_policy; } + + pol = vma->vm_policy; } + if (!pol) + pol = get_task_policy(task); + return pol->flags & MPOL_F_MOF; } @@ -2068,12 +2071,12 @@ retry_cpuset: */ struct page *alloc_pages_current(gfp_t gfp, unsigned order) { - struct mempolicy *pol = get_task_policy(current); + struct mempolicy *pol = &default_policy; struct page *page; unsigned int cpuset_mems_cookie; - if (in_interrupt() || (gfp & __GFP_THISNODE)) - pol = &default_policy; + if (!in_interrupt() && !(gfp & __GFP_THISNODE)) + pol = get_task_policy(current); retry_cpuset: cpuset_mems_cookie = read_mems_allowed_begin(); -- 1.5.5.1