From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S933885Ab1DNSVJ (ORCPT ); Thu, 14 Apr 2011 14:21:09 -0400 Received: from mx1.redhat.com ([209.132.183.28]:18850 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S933271Ab1DNSVD (ORCPT ); Thu, 14 Apr 2011 14:21:03 -0400 Date: Thu, 14 Apr 2011 20:19:36 +0200 From: Oleg Nesterov To: Paul Gortmaker Cc: stable@kernel.org, linux-kernel@vger.kernel.org, stable-review@kernel.org, Linus Torvalds Subject: Re: [34-longterm 136/209] exec: make argv/envp memory visible to oom-killer Message-ID: <20110414181936.GA22589@redhat.com> References: <1302803039-9400-1-git-send-email-paul.gortmaker@windriver.com> <1302803767-9715-1-git-send-email-paul.gortmaker@windriver.com> <1302803767-9715-23-git-send-email-paul.gortmaker@windriver.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1302803767-9715-23-git-send-email-paul.gortmaker@windriver.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 On 04/14, Paul Gortmaker wrote: > > ===================================================================== > | This is a commit scheduled for the next v2.6.34 longterm release. | > | If you see a problem with using this for longterm, please comment.| > ===================================================================== > ... > > With this patch get_arg_page() increments current's MM_ANONPAGES > counter every time we allocate the new page for argv/envp. When > do_execve() succeds or fails, we change this counter back. This only works starting from 2.6.36. before 2.6.36 kernel, oom-killer's badness() uses mm->total_vm. Please see the patch for pre2.6.36 kernel below. Oleg. -------------------------------------------------------------------- commit 3c77f845722158206a7209c45ccddc264d19319c upstream. Brad Spengler published a local memory-allocation DoS that evades the OOM-killer (though not the virtual memory RLIMIT): http://www.grsecurity.net/~spender/64bit_dos.c execve()->copy_strings() can allocate a lot of memory, but this is not visible to oom-killer, nobody can see the nascent bprm->mm and take it into account. With this patch get_arg_page() increments current's MM_ANONPAGES counter every time we allocate the new page for argv/envp. When do_execve() succeds or fails, we change this counter back. Technically this is not 100% correct, we can't know if the new page is swapped out and turn MM_ANONPAGES into MM_SWAPENTS, but I don't think this really matters and everything becomes correct once exec changes ->mm or fails. Compared to upstream: before 2.6.36 kernel, oom-killer's badness() takes mm->total_vm into account and nothing else. So acct_arg_size() has to play with this counter too. Reported-by: Brad Spengler Reviewed-and-discussed-by: KOSAKI Motohiro Signed-off-by: Oleg Nesterov Signed-off-by: Linus Torvalds --- include/linux/binfmts.h | 1 + fs/exec.c | 28 ++++++++++++++++++++++++++-- 2 files changed, 27 insertions(+), 2 deletions(-) --- 2.6.35/include/linux/binfmts.h~1_acct_exec_mem 2010-03-11 13:11:50.000000000 +0100 +++ 2.6.35/include/linux/binfmts.h 2010-12-13 12:01:22.000000000 +0100 @@ -29,6 +29,7 @@ struct linux_binprm{ char buf[BINPRM_BUF_SIZE]; #ifdef CONFIG_MMU struct vm_area_struct *vma; + unsigned long vma_pages; #else # define MAX_ARG_PAGES 32 struct page *page[MAX_ARG_PAGES]; --- 2.6.35/fs/exec.c~1_acct_exec_mem 2010-05-28 13:41:40.000000000 +0200 +++ 2.6.35/fs/exec.c 2010-12-13 12:00:51.000000000 +0100 @@ -158,6 +158,21 @@ out: #ifdef CONFIG_MMU +static void acct_arg_size(struct linux_binprm *bprm, unsigned long pages) +{ + struct mm_struct *mm = current->mm; + long diff = (long)(pages - bprm->vma_pages); + + if (!mm || !diff) + return; + + bprm->vma_pages = pages; + + down_write(&mm->mmap_sem); + mm->total_vm += diff; + up_write(&mm->mmap_sem); +} + static struct page *get_arg_page(struct linux_binprm *bprm, unsigned long pos, int write) { @@ -180,6 +195,8 @@ static struct page *get_arg_page(struct unsigned long size = bprm->vma->vm_end - bprm->vma->vm_start; struct rlimit *rlim; + acct_arg_size(bprm, size / PAGE_SIZE); + /* * We've historically supported up to 32 pages (ARG_MAX) * of argument strings even with small stacks @@ -270,6 +287,10 @@ static bool valid_arg_len(struct linux_b #else +static inline void acct_arg_size(struct linux_binprm *bprm, unsigned long pages) +{ +} + static struct page *get_arg_page(struct linux_binprm *bprm, unsigned long pos, int write) { @@ -977,6 +998,7 @@ int flush_old_exec(struct linux_binprm * /* * Release all of the old mmap stuff */ + acct_arg_size(bprm, 0); retval = exec_mmap(bprm->mm); if (retval) goto out; @@ -1401,8 +1423,10 @@ int do_execve(char * filename, return retval; out: - if (bprm->mm) - mmput (bprm->mm); + if (bprm->mm) { + acct_arg_size(bprm, 0); + mmput(bprm->mm); + } out_file: if (bprm->file) {