From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-fsdevel-owner@vger.kernel.org>
Received: from zeniv.linux.org.uk ([195.92.253.2]:54304 "EHLO
        ZenIV.linux.org.uk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
        with ESMTP id S1753017AbeE1Csb (ORCPT
        <rfc822;linux-fsdevel@vger.kernel.org>);
        Sun, 27 May 2018 22:48:31 -0400
Date: Mon, 28 May 2018 03:48:29 +0100
From: Al Viro <viro@ZenIV.linux.org.uk>
To: Andreas Dilger <adilger@dilger.ca>
Cc: linux-fsdevel@vger.kernel.org,
        Linus Torvalds <torvalds@linux-foundation.org>
Subject: Re: [RFC][PATCH] rmdir(),rename(): do shrink_dcache_parent() only on
 success
Message-ID: <20180528024829.GW30522@ZenIV.linux.org.uk>
References: <20180527222029.GR30522@ZenIV.linux.org.uk>
 <64F20BDA-00AB-429E-959E-1463AE6DCC4F@dilger.ca>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <64F20BDA-00AB-429E-959E-1463AE6DCC4F@dilger.ca>
Sender: linux-fsdevel-owner@vger.kernel.org
List-ID: <linux-fsdevel.vger.kernel.org>

On Sun, May 27, 2018 at 08:14:30PM -0600, Andreas Dilger wrote:

> > 	if (!(flags & RENAME_EXCHANGE) && target) {
> > -		if (is_dir)
> > +		if (is_dir) {
> > +			shrink_dcache_parent(new_dentry);
> > 			target->i_flags |= S_DEAD;
> 
> Would it be better to set S_DEAD on the removed directory before
> shrink_dcache_parent() is called (here and in vfs_rmdir()), or is
> there no way for a new dentry to be added to the parent after the
> shrink is done?

It's locked (exclusive).  Lookups (as well as readdir preseeding
of dcache in case of filesystems that do it) are under the same
lock held at least shared.  The same goes for all IS_DEADDIR callers -
exact same locking is used for S_DEAD handling as well.  So the order
really doesn't matter here.