Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.31 as permitted sender) client-ip=23.128.96.31;
Date:   Fri, 24 Nov 2023 06:15:18 +0000
From:   Al Viro <viro@zeniv.linux.org.uk>
To:     Jan Kara <jack@suse.cz>
Cc:     linux-fsdevel@vger.kernel.org,
        Linus Torvalds <torvalds@linux-foundation.org>,
        Mo Zou <lostzoumo@gmail.com>, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 9/9] rename(): avoid a deadlock in the case of parents
 having no common ancestor
Message-ID: <20231124061518.GS38156@ZenIV>
References: <20231122193028.GE38156@ZenIV>
 <20231122193652.419091-1-viro@zeniv.linux.org.uk>
 <20231122193652.419091-9-viro@zeniv.linux.org.uk>
 <20231123103622.4mfjwrmxr4tl53hi@quack3>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20231123103622.4mfjwrmxr4tl53hi@quack3>
Sender: Al Viro <viro@ftp.linux.org.uk>
Precedence: bulk

On Thu, Nov 23, 2023 at 11:36:22AM +0100, Jan Kara wrote:

> It would look more natural to me if s_vfs_rename_mutex got dropped in the
> callers (lock_rename(), lock_rename_child()) which have acquired the lock
> instead of here. I agree it results in a bit more boiler plate code though.

finish_locking_two_parents(), perhaps? ;-)

Seriously, though - it starts with ->s_vfs_rename_mutex and ends with
the environment for vfs_rename() or with all locks dropped.