From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755044AbZIRCUf (ORCPT ); Thu, 17 Sep 2009 22:20:35 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1753187AbZIRCUe (ORCPT ); Thu, 17 Sep 2009 22:20:34 -0400 Received: from qw-out-2122.google.com ([74.125.92.25]:61714 "EHLO qw-out-2122.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752706AbZIRCUd convert rfc822-to-8bit (ORCPT ); Thu, 17 Sep 2009 22:20:33 -0400 DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type:content-transfer-encoding; b=JknK91HzFTGcMufhafbpJt2Fv3kwgEy1dRrK48x7SSPrlQAG0+cnhFwxInF0fwtWqU htdfdH8otc+slCKywExse95Gpk3w+FhSajsIS5RiFOVdoD0T0EM39JHWk4eh8IEde7t8 uKRH9nkw4uvXRAuoUCr8SZZ9+wiWnbj0NJuU8= MIME-Version: 1.0 In-Reply-To: <20090917151016.99f7c5ab.kamezawa.hiroyu@jp.fujitsu.com> References: <2375c9f90909160235m1f052df0qb001f8243ed9291e@mail.gmail.com> <1bc66b163326564dafb5a7dd8959fd56.squirrel@webmail-b.css.fujitsu.com> <20090917114138.e14a1183.kamezawa.hiroyu@jp.fujitsu.com> <20090917114404.d87b155d.kamezawa.hiroyu@jp.fujitsu.com> <2375c9f90909162302m1fb89414o4f72b6b36e7cbb06@mail.gmail.com> <20090917151016.99f7c5ab.kamezawa.hiroyu@jp.fujitsu.com> Date: Fri, 18 Sep 2009 10:20:36 +0800 Message-ID: <2375c9f90909171920q6941b8al39a045529550732d@mail.gmail.com> Subject: Re: [PATCH 2/3][mmotm] showing size of kcore v2 From: =?UTF-8?Q?Am=C3=A9rico_Wang?= To: KAMEZAWA Hiroyuki Cc: Andrew Morton , linux-kernel@vger.kernel.org, linux-mm@kvack.org Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8BIT Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Sep 17, 2009 at 2:10 PM, KAMEZAWA Hiroyuki wrote: > On Thu, 17 Sep 2009 14:02:39 +0800 > Américo Wang wrote: >> > @@ -124,6 +126,7 @@ static void __kcore_update_ram(struct li >> >        write_unlock(&kclist_lock); >> > >> >        free_kclist_ents(&garbage); >> > +       proc_root_kcore->size = get_kcore_size(&nphdr, &size); >> >> >> This makes me to think if we will have some race condition here? >> Two processes can open kcore at the same time... >> > Finally, > == > static void __kcore_update_ram(struct list_head *list) > { >  write_lock(&kclist_lock); >        if (kcore_need_update) { >                list_for_each_entry_safe(pos, tmp, &kclist_head, list) { >                        if (pos->type == KCORE_RAM >                                || pos->type == KCORE_VMEMMAP) >                                list_move(&pos->list, &garbage); >                } >                list_splice_tail(list, &kclist_head); >        } else >                list_splice(list, &garbage); >        kcore_need_update = 0; >        write_unlock(&kclist_lock); > } > > kclist itself is double checked under write_lock. > And, once updated, get_kcore_size()'s return vaule is static. Imagine one process does get_kcore_size(), then another process is scheduled, who also does get_kcore_size() but at this time, memory size is changed, so it gets a different value. If then the second process writes to proc_root_kcore->size before the first one does, the proc_root_kcore->size is wrong. Am I missing something here? > So, I think there are no race. But..Hmm...is this clearer ? > > == > From: KAMEZAWA Hiroyuki Yes, this version should be OK. Acked-by: WANG Cong > > Now, size of /proc/kcore which can be read by 'ls -l' is 0. > But it's not correct value. > > This is a patch for showing size of /proc/kcore as following. > > On x86-64, ls -l shows >  ... root root 140737486266368 2009-09-17 10:29 /proc/kcore > Then, 7FFFFFFE02000. This comes from vmalloc area's size. > This shows "core" size, not  memory size. > > This patch shows the size by updating "size" field in struct proc_dir_entry. > Later, lookup routine will create inode and fill inode->i_size based > on this value. Then, this has a problem. > >  - Once inode is cached, inode->i_size will never be updated. > > Then, this patch is not memory-hotplug-aware. > > To update inode->i_size, we have to know dentry or inode. > But there is no way to lookup them by inside kernel. Hmmm.... > Next patch will try it. > > Changelog: >  -moved upadting ->size under lock. > > Cc: WANG Cong > Signed-off-by: KAMEZAWA Hiroyuki > --- >  fs/proc/kcore.c |    6 +++++- >  1 file changed, 5 insertions(+), 1 deletion(-) > > Index: mmotm-2.6.31-Sep14/fs/proc/kcore.c > =================================================================== > --- mmotm-2.6.31-Sep14.orig/fs/proc/kcore.c > +++ mmotm-2.6.31-Sep14/fs/proc/kcore.c > @@ -107,6 +107,8 @@ static void free_kclist_ents(struct list >  */ >  static void __kcore_update_ram(struct list_head *list) >  { > +       int nphdr; > +       size_t size; >        struct kcore_list *tmp, *pos; >        LIST_HEAD(garbage); > > @@ -121,6 +123,7 @@ static void __kcore_update_ram(struct li >        } else >                list_splice(list, &garbage); >        kcore_need_update = 0; > +       proc_root_kcore->size = get_kcore_size(&nphdr, &size); >        write_unlock(&kclist_lock); > >        free_kclist_ents(&garbage); > @@ -429,7 +432,8 @@ read_kcore(struct file *file, char __use >        unsigned long start; > >        read_lock(&kclist_lock); > -       proc_root_kcore->size = size = get_kcore_size(&nphdr, &elf_buflen); > +       size = get_kcore_size(&nphdr, &elf_buflen); > + >        if (buflen == 0 || *fpos >= size) { >                read_unlock(&kclist_lock); >                return 0; > > > > > > > >