From mboxrd@z Thu Jan  1 00:00:00 1970
From: Al Viro <viro@ZenIV.linux.org.uk>
Subject: Re: [PATCH 04/11] vfs: check unlinked ancestors before mount
Date: Thu, 5 Sep 2013 14:23:25 +0100
Message-ID: <20130905132325.GQ13318@ZenIV.linux.org.uk>
References: <1378374284-1484-1-git-send-email-miklos@szeredi.hu>
 <1378374284-1484-5-git-send-email-miklos@szeredi.hu>
 <20130905111852.GP13318@ZenIV.linux.org.uk>
 <CAJfpeguzpkw3SYm6kBP-acLSHTgYLXMFRZDf2T0U4aowYhfQdA@mail.gmail.com>
 <20130905120230.GA21170@tucsk.piliscsaba.szeredi.hu>
 <CAJfpegtXQebPEQqw8ESmM-Bp0QgCefmRwoNeJNtN-_bV35b-pw@mail.gmail.com>
 <20130905123911.GA25538@tucsk.piliscsaba.szeredi.hu>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: Linux-Fsdevel <linux-fsdevel@vger.kernel.org>,
	Kernel Mailing List <linux-kernel@vger.kernel.org>,
	"mszeredi@suse.cz" <mszeredi@suse.cz>,
	David Howells <dhowells@redhat.com>,
	Steven Whitehouse <swhiteho@redhat.com>,
	Trond Myklebust <Trond.Myklebust@netapp.com>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: Miklos Szeredi <miklos@szeredi.hu>
Return-path: <linux-kernel-owner@vger.kernel.org>
Content-Disposition: inline
In-Reply-To: <20130905123911.GA25538@tucsk.piliscsaba.szeredi.hu>
Sender: linux-kernel-owner@vger.kernel.org
List-Id: linux-fsdevel.vger.kernel.org

On Thu, Sep 05, 2013 at 02:39:11PM +0200, Miklos Szeredi wrote:
> +static bool __has_unlinked_ancestor(struct dentry *dentry)
> +{
> +	struct dentry *this;
> +
> +	for (this = dentry; !IS_ROOT(this); this = this->d_parent) {
> +		int is_unhashed;
> +
> +		/* Need exclusion wrt. check_submounts_and_drop() */
> +		spin_lock(&this->d_lock);
> +		is_unhashed = d_unhashed(this);
> +		spin_unlock(&this->d_lock);
> +
> +		if (is_unhashed)
> +			return true;
> +	}
> +	return false;
> +}

I still don't get it; why do you need to bother with early setting of
DCACHE_MOUNTED?

You are grabbing rename_lock for write in d_set_mounted().  What kind of races
with check for submounts are you worried about?  d_walk() will rescan
everything if something grabs rename_lock for write while it had been running,
so just fold the "have nothing in d_subdir" case of check_submounts_and_drop()
into d_walk() and be done with that...  What's the problem with such
variant?  AFAICS, all you need to care about is d_set_mounted() not getting
between the scan for submounts and actual __d_drop() and your "finish"
callback is called only after d_walk() having grabbed d_lock *and* rechecked
rename_lock.