Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Wed, 22 Jun 2022 11:34:59 +0100
From:   Al Viro <viro@zeniv.linux.org.uk>
To:     sunliming <sunliming@kylinos.cn>
Cc:     linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
        sunliming@kylino.cn, kelulanainsley@gmail.com
Subject: Re: [PATCH] walk_component(): get inode in lookup_slow branch
 statement block
Message-ID: <YrLwU27DNm0YWOvB@ZenIV>
References: <20220622085146.444516-1-sunliming@kylinos.cn>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220622085146.444516-1-sunliming@kylinos.cn>
Sender: Al Viro <viro@ftp.linux.org.uk>
Precedence: bulk

On Wed, Jun 22, 2022 at 04:51:46PM +0800, sunliming wrote:
> The inode variable is used as a parameter by the step_into function,
> but is not assigned a value in the sub-lookup_slow branch path. So
> get the inode in the sub-lookup_slow branch path.

Take a good look at handle_mounts() and the things it does when
*not* in RCU mode (i.e. LOOKUP_RCU is not set).  Specifically,
                *inode = d_backing_inode(path->dentry);
		*seqp = 0; /* out of RCU mode, so the value doesn't matter */
this part.

IOW, the values passed to step_into() in inode/seq are overridden unless
we stay in RCU mode.  And if we'd been through lookup_slow(), we'd been
out of RCU mode since before we called step_into().