From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932277AbcHCNq7 (ORCPT ); Wed, 3 Aug 2016 09:46:59 -0400 Received: from mail-qt0-f195.google.com ([209.85.216.195]:35123 "EHLO mail-qt0-f195.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754609AbcHCNq4 (ORCPT ); Wed, 3 Aug 2016 09:46:56 -0400 Message-ID: <1470232012.18285.4.camel@poochiereds.net> Subject: Re: [PATCH v2] locks: Filter /proc/locks output on proc pid ns From: Jeff Layton To: Nikolay Borisov , bfields@fieldses.org Cc: viro@zeniv.linux.org.uk, linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org, ebiederm@xmission.com, containers@lists.linux-foundation.org Date: Wed, 03 Aug 2016 09:46:52 -0400 In-Reply-To: <1470209710-30022-1-git-send-email-kernel@kyup.com> References: <1470148943-21835-1-git-send-email-kernel@kyup.com> <1470209710-30022-1-git-send-email-kernel@kyup.com> Content-Type: text/plain; charset="UTF-8" X-Mailer: Evolution 3.20.4 (3.20.4-1.fc24) Mime-Version: 1.0 Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, 2016-08-03 at 10:35 +0300, Nikolay Borisov wrote: > On busy container servers reading /proc/locks shows all the locks > created by all clients. This can cause large latency spikes. In my > case I observed lsof taking up to 5-10 seconds while processing around > 50k locks. Fix this by limiting the locks shown only to those created > in the same pidns as the one the proc was mounted in. When reading > /proc/locks from the init_pid_ns show everything. > > > Signed-off-by: Nikolay Borisov > --- >  fs/locks.c | 6 ++++++ >  1 file changed, 6 insertions(+) > > diff --git a/fs/locks.c b/fs/locks.c > index ee1b15f6fc13..751673d7f7fc 100644 > --- a/fs/locks.c > +++ b/fs/locks.c > @@ -2648,9 +2648,15 @@ static int locks_show(struct seq_file *f, void *v) >  { > >   struct locks_iterator *iter = f->private; > >   struct file_lock *fl, *bfl; > > + struct pid_namespace *proc_pidns = file_inode(f->file)->i_sb->s_fs_info; > > + struct pid_namespace *current_pidns = task_active_pid_ns(current); >   > >   fl = hlist_entry(v, struct file_lock, fl_link); >   > > > + if ((current_pidns != &init_pid_ns) && fl->fl_nspid Ok, so when you read from a process that's in the init_pid_ns namespace, then you'll get the whole pile of locks, even when reading this from a filesystem that was mounted in a different pid_ns? That seems odd to me if so. Any reason not to just uniformly use the proc_pidns here? > > > +     && (proc_pidns != ns_of_pid(fl->fl_nspid))) > > + return 0; > + > >   lock_get_status(f, fl, iter->li_pos, ""); >   > >   list_for_each_entry(bfl, &fl->fl_block, fl_block) -- Jeff Layton