Received-SPF: pass (google.com: domain of linux-nfs+bounces-3619-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Date: Sat, 8 Jun 2024 17:26:06 +0300
From: Dan Carpenter <dan.carpenter@linaro.org>
To: Trond Myklebust <trond.myklebust@hammerspace.com>
Cc: linux-nfs@vger.kernel.org
Subject: [bug report] NFS: Use of mapping_set_error() results in spurious
 errors
Message-ID: <ba14adf7-722e-4281-8746-36311ab01b18@moroto.mountain>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline

Hello Trond Myklebust,

Commit 6c984083ec24 ("NFS: Use of mapping_set_error() results in
spurious errors") from Feb 15, 2022 (linux-next), leads to the
following Smatch static checker warning:

	fs/nfs/write.c:318 nfs_mapping_set_error()
	error: we previously assumed 'mapping->host' could be null (see line 315)

fs/nfs/write.c
    310 static void nfs_mapping_set_error(struct folio *folio, int error)
    311 {
    312         struct address_space *mapping = folio_file_mapping(folio);
    313 
    314         filemap_set_wb_err(mapping, error);
    315         if (mapping->host)
                    ^^^^^^^^^^^^^
The patch adds a check

    316                 errseq_set(&mapping->host->i_sb->s_wb_err,
    317                            error == -ENOSPC ? -ENOSPC : -EIO);
--> 318         nfs_set_pageerror(mapping);
                                  ^^^^^^^
But the next line dereferences mapping->host without checking.

    319 }

regards,
dan carpenter